diff --git a/android/app/build.gradle.kts b/android/app/build.gradle.kts
index 542cf24d8..8c371408e 100644
--- a/android/app/build.gradle.kts
+++ b/android/app/build.gradle.kts
@@ -34,6 +34,25 @@ android {
jvmTarget = JavaVersion.VERSION_11.toString()
}
+ // 添加lint选项
+ lintOptions {
+ isCheckReleaseBuilds = false
+ }
+
+ // 添加packaging配置，排除冲突文件
+ packaging {
+ resources {
+ excludes.add("META-INF/DEPENDENCIES")
+ excludes.add("META-INF/LICENSE")
+ excludes.add("META-INF/LICENSE.txt")
+ excludes.add("META-INF/license.txt")
+ excludes.add("META-INF/NOTICE")
+ excludes.add("META-INF/NOTICE.txt")
+ excludes.add("META-INF/notice.txt")
+ excludes.add("META-INF/*.kotlin_module")
+ }
+ }
+
defaultConfig {
// TODO: Specify your own unique Application ID (https://developer.android.com/studio/build/application-id.html).
applicationId = "com.yunqiinnovation.deepsound"
@@ -82,14 +101,19 @@ android {
dependencies {
+ implementation("io.modelcontextprotocol:kotlin-sdk:0.4.0")
+
+
+ // 添加本地插件模块依赖
+ implementation(project(":azure_speech"))
+ implementation(project(":open_ai_service"))
+ implementation(project(":volcano_speech"))
// 添加OkHttp依赖
implementation("com.squareup.okhttp3:okhttp:4.9.3")
// 添加核心库反糖化
coreLibraryDesugaring("com.android.tools:desugar_jdk_libs:2.0.3")
- // Microsoft 语音识别SDK
- implementation("com.microsoft.cognitiveservices.speech:client-sdk:1.42.0")
// 添加androidx.media依赖
implementation("androidx.media:media:1.6.0")
@@ -100,11 +124,7 @@ dependencies {
// 添加 AndroidX Security 加密 SharedPreferences 依赖
implementation("androidx.security:security-crypto:1.1.0-alpha06")
- // // 添加火山语音合成SDK依赖
- // implementation("com.bytedance.speechengine:speechengine_tts_tob:5.4.8")
-
- // 添加火山语音识别SDK依赖
- // implementation("com.bytedance.speechengine:speechengine_tob:0.0.5")
+
}
flutter {
diff --git a/android/app/src/main/AndroidManifest.xml b/android/app/src/main/AndroidManifest.xml
index f120cc1f7..23c380fb0 100644
--- a/android/app/src/main/AndroidManifest.xml
+++ b/android/app/src/main/AndroidManifest.xml
@@ -27,6 +27,15 @@
+
+
+
+
+
+
+
+
= arrayOf("zh-CN", "en-US")): Boolean {
- try {
- FileLogger.d(TAG, "初始化 Azure 语音服务")
-
- // 检查配置是否为空
- if (subscriptionKey.isEmpty() || serviceRegion.isEmpty()) {
- FileLogger.e(TAG, "Azure 配置信息不完整")
- return false
- }
-
- // 释放之前的资源
- dispose()
-
- this.subscriptionKey = subscriptionKey
- this.serviceRegion = serviceRegion
-
- // 设置语言
- if (supportedLanguages.isNotEmpty()) {
- this.supportedLanguages = supportedLanguages
- }
-
- // 根据支持的语言数量决定是否启用自动语言检测
- this.isAutoDetectLanguage = supportedLanguages.size >= 2
-
- // 如果只有一种语言，设置为当前语言
- if (!isAutoDetectLanguage && supportedLanguages.isNotEmpty()) {
- this.currentLanguage = supportedLanguages[0]
- }
-
- // 创建语音配置
- speechConfig = SpeechConfig.fromSubscription(subscriptionKey, serviceRegion)
-
- // 设置语言配置
- if (isAutoDetectLanguage) {
- // 设置自动语言检测
- speechConfig?.setProperty(PropertyId.SpeechServiceConnection_LanguageIdMode, "Continuous")
- } else {
- // 设置指定的识别语言
- speechConfig?.speechRecognitionLanguage = currentLanguage
- }
-
- // 创建识别器
- try {
- if (useEchoCancellation) {
- // 如果使用回音消除，创建自定义音频输入流
- setupCustomAudioProcessing()
-
- if (isAutoDetectLanguage) {
- val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
- recognizer = SpeechRecognizer(speechConfig, autoDetectConfig, audioConfig)
- } else {
- recognizer = SpeechRecognizer(speechConfig, audioConfig)
- }
- } else {
- // 使用默认麦克风输入
- if (isAutoDetectLanguage) {
- val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
- recognizer = SpeechRecognizer(speechConfig, autoDetectConfig)
- } else {
- recognizer = SpeechRecognizer(speechConfig)
- }
- }
-
- FileLogger.d(TAG, "Azure 语音服务初始化成功")
- return true
- } catch (e: Exception) {
- FileLogger.e(TAG, "创建识别器失败: ${e.message}")
- stopCustomAudioProcessing()
- return false
- }
- } catch (e: Exception) {
- FileLogger.e(TAG, "初始化失败: ${e.message}")
- return false
- }
- }
-
- // 重置 recognizer
- private fun resetRecognizer(): Boolean {
- try {
- // 释放之前的 recognizer
- recognizer?.close()
- recognizer = null
-
- // 停止当前的音频处理
- stopCustomAudioProcessing()
-
- // 使用现有配置重新创建 recognizer
- if (speechConfig != null) {
- if (useEchoCancellation) {
- // 如果使用回音消除，创建自定义音频输入流
- setupCustomAudioProcessing()
-
- if (isAutoDetectLanguage) {
- val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
- recognizer = SpeechRecognizer(speechConfig, autoDetectConfig, audioConfig)
- } else {
- recognizer = SpeechRecognizer(speechConfig, audioConfig)
- }
- } else {
- // 使用默认麦克风输入
- if (isAutoDetectLanguage) {
- val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
- recognizer = SpeechRecognizer(speechConfig, autoDetectConfig)
- } else {
- recognizer = SpeechRecognizer(speechConfig)
- }
- }
- return true
- } else {
- FileLogger.e(TAG, "语音配置未初始化")
- return false
- }
- } catch (e: Exception) {
- FileLogger.e(TAG, "重置识别器失败: ${e.message}")
- return false
- }
- }
-
- // 开始一次性语音识别
- fun recognizeOnce(callback: RecognizeCallback) {
- if (speechConfig == null) {
- callback.onError("语音服务未初始化")
- return
- }
-
- // 重置 recognizer
- if (!resetRecognizer()) {
- callback.onError("重置识别器失败")
- return
- }
-
- try {
- // 启动音频处理
- startCustomAudioProcessing()
-
- // 执行识别
- val result = recognizer?.recognizeOnceAsync()?.get()
-
- // 停止音频处理
- stopCustomAudioProcessing()
-
- if (result != null && result.reason == ResultReason.RecognizedSpeech) {
- val detectedLanguage = AutoDetectSourceLanguageResult.fromResult(result)?.language
- callback.onResult(result.text, detectedLanguage ?: "")
- } else {
- callback.onError("未能识别语音")
- }
- } catch (e: Exception) {
- // 停止音频处理
- stopCustomAudioProcessing()
- callback.onError("识别异常: ${e.message}")
- }
- }
-
- // 开始连续语音识别
- fun startContinuousRecognition(callback: ContinuousRecognizeCallback): Boolean {
- if (speechConfig == null) {
- callback.onError("语音服务未初始化")
- return false
- }
-
- // 如果已经在进行连续识别，先停止
- if (isContinuousRecognitionActive) {
- stopContinuousRecognition(callback)
- }
-
- // 重置 recognizer
- if (!resetRecognizer()) {
- callback.onError("重置识别器失败")
- return false
- }
-
- try {
- // 启动音频处理
- startCustomAudioProcessing()
-
- // 设置识别事件处理
- // 最终识别结果
- recognizer?.recognized?.addEventListener(
- EventHandler { _, event ->
- if (event.result.reason == ResultReason.RecognizedSpeech) {
- val detectedLanguage = if (isAutoDetectLanguage) {
- AutoDetectSourceLanguageResult.fromResult(event.result)?.language ?: ""
- } else {
- currentLanguage
- }
- // FileLogger.d(TAG, "最终识别结果: ${event.result.text}")
- callback.onResult(event.result.text, detectedLanguage)
- }
- }
- )
-
- // 识别中事件
- recognizer?.recognizing?.addEventListener(
- EventHandler { _, event ->
- if (event.result.reason == ResultReason.RecognizingSpeech) {
- val detectedLanguage = if (isAutoDetectLanguage) {
- AutoDetectSourceLanguageResult.fromResult(event.result)?.language ?: ""
- } else {
- currentLanguage
- }
- // FileLogger.d(TAG, "识别中结果: ${event.result.text}")
- callback.onRecognizing(event.result.text, detectedLanguage)
- }
- }
- )
-
- // 会话事件
- recognizer?.sessionStarted?.addEventListener(
- EventHandler { _, _ ->
- isContinuousRecognitionActive = true
- callback.onSessionStarted()
- }
- )
-
- recognizer?.sessionStopped?.addEventListener(
- EventHandler { _, _ ->
- isContinuousRecognitionActive = false
- callback.onSessionStopped()
- }
- )
-
- // 取消事件
- recognizer?.canceled?.addEventListener(
- EventHandler { _, event ->
- val errorDetails = if (event.reason == CancellationReason.Error) event.errorDetails else ""
- callback.onCanceled(event.reason.toString(), errorDetails)
- isContinuousRecognitionActive = false
- }
- )
-
- // 开始连续识别
- recognizer?.startContinuousRecognitionAsync()?.get()
- isContinuousRecognitionActive = true
-
- return true
- } catch (e: Exception) {
- callback.onError("开始连续识别失败: ${e.message}")
- isContinuousRecognitionActive = false
- stopCustomAudioProcessing()
- return false
- }
- }
-
- // 停止连续语音识别
- fun stopContinuousRecognition(callback: ContinuousRecognizeCallback): Boolean {
- if (!isContinuousRecognitionActive || recognizer == null) {
- return true
- }
-
- try {
- recognizer?.stopContinuousRecognitionAsync()
- isContinuousRecognitionActive = false
- callback.onSessionStopped()
-
- // 停止音频处理
- stopCustomAudioProcessing()
-
- return true
- } catch (e: Exception) {
- callback.onError("停止连续识别失败: ${e.message}")
- isContinuousRecognitionActive = false
- stopCustomAudioProcessing()
- return false
- }
- }
-
- // 检查连续识别是否活跃
- fun isContinuousRecognitionActive(): Boolean {
- return isContinuousRecognitionActive
- }
-
- // 设置自定义音频处理
- private fun setupCustomAudioProcessing() {
- if (!useEchoCancellation) {
- return
- }
-
- try {
- // 1. 创建PushAudioInputStream
- pushStream = PushAudioInputStream.create()
-
- // 2. 创建AudioConfig
- audioConfig = AudioConfig.fromStreamInput(pushStream)
-
- // 3. 创建自定义音频处理器
- customAudioProcessor = CustomAudioProcessor(pushStream)
-
- FileLogger.d(TAG, "自定义音频处理设置完成")
- } catch (e: Exception) {
- FileLogger.e(TAG, "设置自定义音频处理失败: ${e.message}")
- releaseCustomAudioProcessing()
-
- // 降级处理：如果自定义处理设置失败，尝试使用默认麦克风
- try {
- FileLogger.d(TAG, "尝试降级到默认麦克风输入")
- audioConfig = AudioConfig.fromDefaultMicrophoneInput()
- } catch (e2: Exception) {
- FileLogger.e(TAG, "默认麦克风输入设置也失败: ${e2.message}")
- audioConfig = null
- }
- }
- }
-
- // 启动自定义音频处理
- private fun startCustomAudioProcessing() {
- if (!useEchoCancellation || customAudioProcessor == null) {
- return
- }
-
- try {
- customAudioProcessor?.startRecording()
- FileLogger.d(TAG, "自定义音频处理已启动")
- } catch (e: Exception) {
- FileLogger.e(TAG, "启动自定义音频处理失败: ${e.message}")
- }
- }
-
- // 停止自定义音频处理
- private fun stopCustomAudioProcessing() {
- if (!useEchoCancellation || customAudioProcessor == null) {
- return
- }
-
- try {
- customAudioProcessor?.stopRecording()
- FileLogger.d(TAG, "自定义音频处理已停止")
- } catch (e: Exception) {
- FileLogger.e(TAG, "停止自定义音频处理失败: ${e.message}")
- }
- }
-
- // 释放自定义音频处理资源
- private fun releaseCustomAudioProcessing() {
- stopCustomAudioProcessing()
-
- try {
- customAudioProcessor = null
- pushStream?.close()
- pushStream = null
- audioConfig?.close()
- audioConfig = null
-
- FileLogger.d(TAG, "自定义音频处理资源已释放")
- } catch (e: Exception) {
- FileLogger.e(TAG, "释放自定义音频处理资源时出错: ${e.message}")
- }
- }
-
- // 释放所有资源
- fun dispose() {
- try {
- // 停止和释放音频处理
- releaseCustomAudioProcessing()
-
- recognizer?.close()
- recognizer = null
-
- speechConfig?.close()
- speechConfig = null
-
- isContinuousRecognitionActive = false
-
- FileLogger.d(TAG, "语音识别资源已释放")
- } catch (e: Exception) {
- FileLogger.e(TAG, "释放资源出错: ${e.message}")
- }
- }
-
- // 自定义音频处理器 - 使用Android原生回音消除
- private inner class CustomAudioProcessor(private val pushStream: PushAudioInputStream?) {
- private val SAMPLE_RATE = 16000
- private val CHANNEL_CONFIG = AudioFormat.CHANNEL_IN_MONO
- private val AUDIO_FORMAT = AudioFormat.ENCODING_PCM_16BIT
- private val BUFFER_SIZE = SAMPLE_RATE * 2 // 简化缓冲区大小计算，更稳定
-
- private var audioRecord: AudioRecord? = null
- private var echoCanceler: AcousticEchoCanceler? = null
- private val isRecording = AtomicBoolean(false)
- private var recordingThread: Thread? = null
-
- // 启动录音并处理音频数据
- fun startRecording() {
- if (isRecording.get() || pushStream == null) {
- return
- }
-
- try {
- // 使用Builder模式构建AudioFormat
- val audioFormat = AudioFormat.Builder()
- .setSampleRate(SAMPLE_RATE)
- .setEncoding(AUDIO_FORMAT)
- .setChannelMask(CHANNEL_CONFIG)
- .build()
-
- // 使用Builder模式创建AudioRecord实例
- audioRecord = AudioRecord.Builder()
- .setAudioSource(MediaRecorder.AudioSource.VOICE_COMMUNICATION)
- .setAudioFormat(audioFormat)
- .setBufferSizeInBytes(BUFFER_SIZE)
- .build()
-
- // 检查AudioRecord初始化状态
- if (audioRecord?.state != AudioRecord.STATE_INITIALIZED) {
- FileLogger.e(TAG, "AudioRecord初始化失败，状态: ${audioRecord?.state}")
- // 尝试使用DEFAULT音频源重试一次
- audioRecord?.release()
- audioRecord = AudioRecord.Builder()
- .setAudioSource(MediaRecorder.AudioSource.DEFAULT)
- .setAudioFormat(audioFormat)
- .setBufferSizeInBytes(BUFFER_SIZE)
- .build()
-
- if (audioRecord?.state != AudioRecord.STATE_INITIALIZED) {
- FileLogger.e(TAG, "AudioRecord初始化第二次尝试也失败，放弃")
- releaseAudioResources()
- return
- } else {
- FileLogger.d(TAG, "使用默认音频源成功初始化AudioRecord")
- }
- }
-
- // 启用音频效果（回音消除、噪声抑制等）
- enableAudioEffects()
-
- // 启动录音
- audioRecord?.startRecording()
- isRecording.set(true)
-
- // 创建录音线程
- recordingThread = Thread({
- val buffer = ByteArray(BUFFER_SIZE)
-
- while (isRecording.get()) {
- try {
- val readSize = audioRecord?.read(buffer, 0, BUFFER_SIZE) ?: 0
-
- if (readSize > 0) {
- try {
- // 将处理后的音频数据推送到流
- if (readSize == buffer.size) {
- // 如果读取的大小等于buffer的大小，直接写入整个buffer
- pushStream.write(buffer)
- } else {
- // 如果只读取了部分数据，创建新的数组只包含有效数据
- val validData = buffer.copyOfRange(0, readSize)
- pushStream.write(validData)
- }
- } catch (e: Exception) {
- FileLogger.e(TAG, "写入音频数据失败: ${e.message}")
- break
- }
- } else if (readSize == 0) {
- // 读取为0，可能是临时的，等待一下继续尝试
- Thread.sleep(10)
- } else {
- // 负值表示错误
- FileLogger.e(TAG, "读取音频数据失败，错误码: $readSize")
- break
- }
- } catch (e: Exception) {
- FileLogger.e(TAG, "录音线程异常: ${e.message}")
- break
- }
- }
- }, "AudioRecordingThread")
-
- // 设置线程优先级并启动
- recordingThread?.priority = Thread.MAX_PRIORITY
- recordingThread?.start()
-
- FileLogger.d(TAG, "音频录制已启动" + (if(echoCanceler?.enabled == true) "，回音消除已启用" else ""))
- } catch (e: Exception) {
- FileLogger.e(TAG, "启动音频录制失败: ${e.message}")
- releaseAudioResources()
- }
- }
-
- // 启用音频效果（回音消除、噪声抑制等）
- private fun enableAudioEffects() {
- try {
- val audioSessionId = audioRecord?.audioSessionId ?: -1
-
- if (audioSessionId != -1) {
- // 启用回音消除
- if (AcousticEchoCanceler.isAvailable()) {
- try {
- echoCanceler = AcousticEchoCanceler.create(audioSessionId)
- if (echoCanceler != null) {
- echoCanceler?.enabled = true
- FileLogger.d(TAG, "回音消除已启用，会话ID: $audioSessionId")
- } else {
- FileLogger.w(TAG, "回音消除器创建返回null")
- }
- } catch (e: Exception) {
- FileLogger.e(TAG, "创建回音消除器时出错: ${e.message}")
- }
- } else {
- FileLogger.d(TAG, "设备不支持回音消除")
- }
-
- // 以下功能暂时不启用，可根据需要取消注释
- /*
- // 启用噪声抑制
- if (NoiseSuppressor.isAvailable()) {
- try {
- val ns = NoiseSuppressor.create(audioSessionId)
- ns?.enabled = true
- FileLogger.d(TAG, "噪声抑制已启用")
- } catch (e: Exception) {
- FileLogger.e(TAG, "创建噪声抑制器时出错: ${e.message}")
- }
- }
-
- // 启用自动增益控制
- if (AutomaticGainControl.isAvailable()) {
- try {
- val agc = AutomaticGainControl.create(audioSessionId)
- agc?.enabled = true
- FileLogger.d(TAG, "自动增益控制已启用")
- } catch (e: Exception) {
- FileLogger.e(TAG, "创建自动增益控制时出错: ${e.message}")
- }
- }
- */
- } else {
- FileLogger.w(TAG, "无效的音频会话ID，无法启用音频效果")
- }
- } catch (e: Exception) {
- FileLogger.e(TAG, "启用音频效果时出错: ${e.message}")
- }
- }
-
- // 停止录音
- fun stopRecording() {
- if (!isRecording.get()) {
- return
- }
-
- isRecording.set(false)
-
- try {
- // 等待录音线程结束
- recordingThread?.join(1000)
-
- // 释放资源
- releaseAudioResources()
-
- FileLogger.d(TAG, "音频录制已停止")
- } catch (e: Exception) {
- FileLogger.e(TAG, "停止音频录制失败: ${e.message}")
- }
- }
-
- // 释放音频资源
- private fun releaseAudioResources() {
- try {
- // 停止录音
- try {
- if (audioRecord?.state == AudioRecord.STATE_INITIALIZED) {
- audioRecord?.stop()
- }
- } catch (e: Exception) {
- // 忽略可能的IllegalStateException
- FileLogger.w(TAG, "停止AudioRecord时出错: ${e.message}")
- }
-
- // 释放回音消除器
- try {
- if (echoCanceler != null) {
- echoCanceler?.enabled = false
- echoCanceler?.release()
- echoCanceler = null
- }
- } catch (e: Exception) {
- FileLogger.w(TAG, "释放回音消除器时出错: ${e.message}")
- } finally {
- echoCanceler = null
- }
-
- // 释放音频记录器
- try {
- audioRecord?.release()
- } catch (e: Exception) {
- FileLogger.w(TAG, "释放AudioRecord时出错: ${e.message}")
- } finally {
- audioRecord = null
- }
-
- // 重置线程
- recordingThread = null
-
- } catch (e: Exception) {
- FileLogger.e(TAG, "释放音频资源失败: ${e.message}")
- }
- }
- }
-
- // 一次性识别回调接口
- interface RecognizeCallback {
- fun onResult(result: String, detectedLanguage: String = "")
- fun onError(error: String)
- }
-
- // 连续识别回调接口
- interface ContinuousRecognizeCallback {
- fun onResult(result: String, detectedLanguage: String = "")
- fun onRecognizing(recognizing: String, detectedLanguage: String = "")
- fun onSessionStarted()
- fun onSessionStopped()
- fun onCanceled(reason: String, errorDetails: String)
- fun onError(error: String)
- }
-}
\ No newline at end of file
diff --git a/android/app/src/main/kotlin/com/example/deep_voice/VolcanoAIService.kt b/android/app/src/main/kotlin/com/example/deep_voice/VolcanoAIService.kt
deleted file mode 100644
index 2fbcd8da8..000000000
--- a/android/app/src/main/kotlin/com/example/deep_voice/VolcanoAIService.kt
+++ /dev/null
@@ -1,344 +0,0 @@
-package com.yunqiinnovation.deepsound
-
-import android.util.Log
-import okhttp3.*
-import okhttp3.MediaType.Companion.toMediaTypeOrNull
-import okhttp3.RequestBody.Companion.toRequestBody
-import org.json.JSONArray
-import org.json.JSONObject
-import java.io.IOException
-import java.util.concurrent.CountDownLatch
-import java.util.concurrent.TimeUnit
-
-/**
- * 火山AI服务的原生实现
- *
- * 参考Flutter端的VolcanoAIService实现，提供同步和异步的API调用方式
- */
-class VolcanoAIService() {
- private val TAG = "VolcanoAIService"
- private val baseUrl = "https://ark.cn-beijing.volces.com/api/v3"
- private val chatEndpoint = "/chat/completions"
- private val client = OkHttpClient.Builder()
- .connectTimeout(30, TimeUnit.SECONDS)
- .readTimeout(30, TimeUnit.SECONDS)
- .writeTimeout(30, TimeUnit.SECONDS)
- .build()
-
- private var apiKey: String = ""
- private var isInitialized = false
-
- /**
- * 初始化火山AI服务
- *
- * @param apiKey 火山AI API密钥
- * @return 初始化是否成功
- */
- fun initialize(apiKey: String): Boolean {
- this.apiKey = apiKey
- isInitialized = apiKey.isNotEmpty()
-
- if (!isInitialized) {
- Log.e(TAG, "初始化失败：API key 不能为空")
- } else {
- Log.d(TAG, "火山AI服务初始化成功")
- }
-
- return isInitialized
- }
-
- /**
- * 生成个性化问候语
- *
- * @param agentName 代理名称
- * @param systemPrompt 系统提示词
- * @param callback 回调函数，返回生成的问候语
- */
- fun generateGreeting(agentName: String, systemPrompt: String, callback: (String?, Exception?) -> Unit) {
- val messages = JSONArray().apply {
- put(JSONObject().apply {
- put("role", "system")
- put("content", systemPrompt)
- })
- put(JSONObject().apply {
- put("role", "user")
- put("content", "请用一句简短的话向我打个招呼，要符合你的身份和性格特点，不要超过18个字。")
- })
- }
-
- sendMessageStream(messages, systemPrompt, object : StreamCallback {
- val stringBuilder = StringBuilder()
-
- override fun onToken(token: String) {
- stringBuilder.append(token)
- }
-
- override fun onComplete() {
- callback(stringBuilder.toString(), null)
- }
-
- override fun onError(e: Exception) {
- callback(null, e)
- }
- })
- }
-
- /**
- * 发送消息（非流式输出）
- *
- * @param messages 消息列表
- * @param systemPrompt 系统提示词
- * @return 返回AI的回复
- * @throws VolcanoAIException 如果API调用失败
- */
- @Throws(VolcanoAIException::class)
- fun sendMessage(messages: JSONArray, systemPrompt: String): String {
- // 检查是否已初始化
- if (!isInitialized || apiKey.isEmpty()) {
- throw VolcanoAIException("火山AI服务未初始化或API key为空，请先调用initialize方法")
- }
-
- val fullMessages = JSONArray().apply {
- put(JSONObject().apply {
- put("role", "system")
- put("content", systemPrompt)
- })
- for (i in 0 until messages.length()) {
- put(messages.getJSONObject(i))
- }
- }
-
- val requestBody = JSONObject().apply {
- put("model", "doubao-1-5-lite-32k-250115")
- put("messages", fullMessages)
- put("temperature", 0.7)
- put("max_tokens", 2000)
- put("stream", false)
- }
-
- val mediaType = "application/json".toMediaTypeOrNull()
- val request = Request.Builder()
- .url("$baseUrl$chatEndpoint")
- .addHeader("Content-Type", "application/json")
- .addHeader("Authorization", "Bearer $apiKey")
- .post(requestBody.toString().toRequestBody(mediaType))
- .build()
-
- try {
- client.newCall(request).execute().use { response ->
- if (!response.isSuccessful) {
- val errorBody = response.body?.string() ?: ""
- val errorMessage = try {
- JSONObject(errorBody).getJSONObject("error").getString("message")
- } catch (e: Exception) {
- "Unknown error occurred"
- }
- throw VolcanoAIException(errorMessage)
- }
-
- val responseBody = response.body?.string() ?: throw VolcanoAIException("Empty response")
- val jsonResponse = JSONObject(responseBody)
-
- if (jsonResponse.has("choices") &&
- jsonResponse.getJSONArray("choices").length() > 0 &&
- jsonResponse.getJSONArray("choices").getJSONObject(0).has("message")) {
- return jsonResponse.getJSONArray("choices")
- .getJSONObject(0)
- .getJSONObject("message")
- .getString("content")
- }
-
- throw VolcanoAIException("Invalid response format")
- }
- } catch (e: Exception) {
- if (e is VolcanoAIException) throw e
- throw VolcanoAIException("Failed to communicate with AI service: ${e.message}")
- }
- }
-
- /**
- * 发送消息（流式输出）
- *
- * @param messages 消息列表
- * @param systemPrompt 系统提示词
- * @param callback 回调函数，用于接收流式输出的结果
- */
- fun sendMessageStream(messages: JSONArray, systemPrompt: String, callback: StreamCallback) {
- // 检查是否已初始化
- if (!isInitialized || apiKey.isEmpty()) {
- callback.onError(VolcanoAIException("火山AI服务未初始化或API key为空，请先调用initialize方法"))
- return
- }
-
- val fullMessages = JSONArray().apply {
- put(JSONObject().apply {
- put("role", "system")
- put("content", systemPrompt)
- })
- for (i in 0 until messages.length()) {
- put(messages.getJSONObject(i))
- }
- }
-
- val requestBody = JSONObject().apply {
- put("model", "doubao-1-5-lite-32k-250115")
- put("messages", fullMessages)
- put("temperature", 0.7)
- put("max_tokens", 2000)
- put("stream", true)
- }
-
- val mediaType = "application/json".toMediaTypeOrNull()
- val request = Request.Builder()
- .url("$baseUrl$chatEndpoint")
- .addHeader("Content-Type", "application/json")
- .addHeader("Authorization", "Bearer $apiKey")
- .addHeader("Accept", "text/event-stream")
- .post(requestBody.toString().toRequestBody(mediaType))
- .build()
-
- client.newCall(request).enqueue(object : Callback {
- override fun onFailure(call: Call, e: IOException) {
- callback.onError(VolcanoAIException("Failed to communicate with AI service: ${e.message}"))
- }
-
- override fun onResponse(call: Call, response: Response) {
- if (!response.isSuccessful) {
- val errorBody = response.body?.string() ?: ""
- val errorMessage = try {
- JSONObject(errorBody).getJSONObject("error").getString("message")
- } catch (e: Exception) {
- "Unknown error occurred"
- }
- callback.onError(VolcanoAIException(errorMessage))
- return
- }
-
- val responseBody = response.body ?: return
- val source = responseBody.source()
- val bufferedSource = source.buffer
-
- try {
- while (!bufferedSource.exhausted()) {
- val line = bufferedSource.readUtf8Line() ?: continue
-
- if (line.isEmpty()) continue
- if (line.startsWith("data: ")) {
- val data = line.substring(6)
- if (data == "[DONE]") {
- callback.onComplete()
- break
- }
-
- try {
- val jsonData = JSONObject(data)
- if (jsonData.has("choices") &&
- jsonData.getJSONArray("choices").length() > 0 &&
- jsonData.getJSONArray("choices").getJSONObject(0).has("delta") &&
- jsonData.getJSONArray("choices").getJSONObject(0).getJSONObject("delta").has("content")) {
- val content = jsonData.getJSONArray("choices")
- .getJSONObject(0)
- .getJSONObject("delta")
- .getString("content")
- callback.onToken(content)
- }
- } catch (e: Exception) {
- // 忽略无效的JSON数据
- continue
- }
- }
- }
- } catch (e: Exception) {
- callback.onError(VolcanoAIException("Error processing stream: ${e.message}"))
- } finally {
- response.close()
- }
- }
- })
- }
-
- /**
- * 同步方式发送消息（流式输出）
- *
- * 注意：此方法会阻塞当前线程，请在后台线程中调用
- *
- * @param messages 消息列表
- * @param systemPrompt 系统提示词
- * @return 返回完整的AI回复
- * @throws VolcanoAIException 如果API调用失败
- */
- @Throws(VolcanoAIException::class)
- fun sendMessageStreamSync(messages: JSONArray, systemPrompt: String): String {
- val result = StringBuilder()
- val latch = CountDownLatch(1)
- var exception: Exception? = null
-
- sendMessageStream(messages, systemPrompt, object : StreamCallback {
- override fun onToken(token: String) {
- result.append(token)
- }
-
- override fun onComplete() {
- latch.countDown()
- }
-
- override fun onError(e: Exception) {
- exception = e
- latch.countDown()
- }
- })
-
- // 等待流式输出完成或出错
- latch.await(60, TimeUnit.SECONDS)
-
- if (exception != null) {
- throw exception as VolcanoAIException
- }
-
- return result.toString()
- }
-
- /**
- * 创建用户消息
- */
- fun createUserMessage(content: String): JSONObject {
- return JSONObject().apply {
- put("role", "user")
- put("content", content)
- }
- }
-
- /**
- * 创建系统消息
- */
- fun createSystemMessage(content: String): JSONObject {
- return JSONObject().apply {
- put("role", "system")
- put("content", content)
- }
- }
-
- /**
- * 创建助手消息
- */
- fun createAssistantMessage(content: String): JSONObject {
- return JSONObject().apply {
- put("role", "assistant")
- put("content", content)
- }
- }
-
- /**
- * 流式输出回调接口
- */
- interface StreamCallback {
- fun onToken(token: String)
- fun onComplete()
- fun onError(e: Exception)
- }
-}
-
-/**
- * 火山AI异常
- */
-class VolcanoAIException(message: String) : Exception(message)
\ No newline at end of file
diff --git a/android/app/src/main/kotlin/com/example/deep_voice/ClassicBluetoothHelper.kt b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/ClassicBluetoothHelper.kt
similarity index 100%
rename from android/app/src/main/kotlin/com/example/deep_voice/ClassicBluetoothHelper.kt
rename to android/app/src/main/kotlin/com/yunqiinnovation/deepsound/ClassicBluetoothHelper.kt
diff --git a/android/app/src/main/kotlin/com/example/deep_voice/MainActivity.kt b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/MainActivity.kt
similarity index 67%
rename from android/app/src/main/kotlin/com/example/deep_voice/MainActivity.kt
rename to android/app/src/main/kotlin/com/yunqiinnovation/deepsound/MainActivity.kt
index 49fd8fdf6..1c4717f11 100644
--- a/android/app/src/main/kotlin/com/example/deep_voice/MainActivity.kt
+++ b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/MainActivity.kt
@@ -1,5 +1,6 @@
package com.yunqiinnovation.deepsound
+import android.Manifest
import android.content.Intent
import android.os.Build
import android.os.Bundle
@@ -20,33 +21,54 @@ import android.bluetooth.BluetoothDevice
import androidx.security.crypto.EncryptedSharedPreferences
import androidx.security.crypto.MasterKey
import com.yunqiinnovation.deepsound.core.utils.FileLogger
+import android.content.pm.PackageManager
+import androidx.core.app.ActivityCompat
+import androidx.core.content.ContextCompat
class MainActivity: FlutterActivity() {
- private val AZURE_ASR_CHANNEL = "com.deep_voice.azure_asr"
- private val AZURE_ASR_EVENT_CHANNEL = "com.deep_voice.azure_asr_events"
- private val AZURE_TTS_CHANNEL = "com.deep_voice.azure_tts"
private val VOICE_INTERACTION_CHANNEL = "com.deep_voice.voice_interaction"
private val VOICE_INTERACTION_EVENT_CHANNEL = "com.deep_voice.voice_interaction_events"
private val CLASSIC_BLUETOOTH_CHANNEL = "com.deep_voice.classic_bluetooth"
private val CLASSIC_BLUETOOTH_EVENT_CHANNEL = "com.deep_voice.classic_bluetooth_events"
private val TAG = "MainActivity"
- private lateinit var azureAsrHelper: AzureAsrHelper
- private lateinit var azureTtsHelper: AzureTtsHelper
private lateinit var classicBluetoothHelper: ClassicBluetoothHelper
- private var azureAsrEventSink: EventChannel.EventSink? = null
private var voiceInteractionEventSink: EventChannel.EventSink? = null
private var bluetoothEventSink: EventChannel.EventSink? = null
+ // 添加权限请求相关常量
+ private val PERMISSION_REQUEST_CODE = 100
+ private val REQUIRED_PERMISSIONS = if (Build.VERSION.SDK_INT >= Build.VERSION_CODES.S) {
+ arrayOf(
+ Manifest.permission.RECORD_AUDIO,
+ Manifest.permission.BLUETOOTH_CONNECT,
+ Manifest.permission.BLUETOOTH_SCAN,
+ Manifest.permission.ACCESS_FINE_LOCATION,
+ Manifest.permission.SEND_SMS,
+ Manifest.permission.READ_CONTACTS,
+ Manifest.permission.CALL_PHONE,
+ Manifest.permission.POST_NOTIFICATIONS
+ )
+ } else {
+ arrayOf(
+ Manifest.permission.RECORD_AUDIO,
+ Manifest.permission.BLUETOOTH,
+ Manifest.permission.BLUETOOTH_ADMIN,
+ Manifest.permission.ACCESS_FINE_LOCATION,
+ Manifest.permission.SEND_SMS,
+ Manifest.permission.READ_CONTACTS,
+ Manifest.permission.CALL_PHONE
+ )
+ }
+
// 广播接收器
private val voiceInteractionReceiver = object : BroadcastReceiver() {
override fun onReceive(context: Context?, intent: Intent?) {
- Log.d(TAG, "收到广播: ${intent?.action}")
+ FileLogger.d(TAG, "收到广播: ${intent?.action}")
when (intent?.action) {
VoiceInteractionService.ACTION_RECOGNITION_STARTED -> {
val timestamp = intent.getLongExtra("timestamp", 0)
- Log.d(TAG, "收到语音识别启动广播: timestamp=$timestamp")
sendVoiceInteractionEvent(mapOf(
"type" to "recognition_started",
"timestamp" to timestamp
@@ -58,12 +80,14 @@ class MainActivity: FlutterActivity() {
val assistantMessage = intent.getStringExtra("assistantMessage") ?: ""
val timestamp = intent.getLongExtra("timestamp", System.currentTimeMillis())
- Log.d(TAG, "收到聊天记录更新广播: agentId=$agentId, timestamp=$timestamp")
- Log.d(TAG, "用户消息: ${userMessage.take(50)}...")
- Log.d(TAG, "助手回复: ${assistantMessage.take(50)}...")
-
sendChatHistoryEvent(agentId, userMessage, assistantMessage, timestamp)
}
+ VoiceInteractionService.ACTION_ENTER_TRANSLATION_MODE -> {
+ sendVoiceInteractionEvent(mapOf(
+ "type" to "enter_translation_mode",
+ "timestamp" to System.currentTimeMillis()
+ ))
+ }
}
}
}
@@ -143,13 +167,24 @@ class MainActivity: FlutterActivity() {
var azureSpeechKey: String = ""
var azureSpeechRegion: String = ""
var volcanoAiApiKey: String = ""
+ var openaiApiKey: String = ""
+ var openaiBaseUrl: String? = null
+ var openaiModel: String? = null
+ var volcanoSpeechAppId: String = ""
+ var volcanoSpeechAppToken: String = ""
// 安全存储相关常量
private const val SECURE_PREFS_FILENAME = "deep_voice_secure_prefs"
private const val KEY_AZURE_SPEECH_KEY = "azure_speech_key"
private const val KEY_AZURE_SPEECH_REGION = "azure_speech_region"
private const val KEY_VOLCANO_AI_API_KEY = "volcano_ai_api_key"
-
+ private const val KEY_OPENAI_API_KEY = "openai_api_key"
+ private const val KEY_OPENAI_BASE_URL = "openai_base_url"
+ private const val KEY_OPENAI_MODEL = "openai_model"
+ private const val KEY_MCP_SERVER_ENDPOINT = "mcp_server_endpoint"
+ private const val KEY_VOLCANO_SPEECH_APP_ID = "volcano_speech_app_id"
+ private const val KEY_VOLCANO_SPEECH_APP_TOKEN = "volcano_speech_app_token"
+
// 会话管理
private const val KEY_SESSION_ID = "session_id"
private var currentSessionId = ""
@@ -183,7 +218,12 @@ class MainActivity: FlutterActivity() {
.putString(KEY_AZURE_SPEECH_KEY, azureSpeechKey)
.putString(KEY_AZURE_SPEECH_REGION, azureSpeechRegion)
.putString(KEY_VOLCANO_AI_API_KEY, volcanoAiApiKey)
+ .putString(KEY_OPENAI_API_KEY, openaiApiKey)
+ .putString(KEY_OPENAI_BASE_URL, openaiBaseUrl)
.putString(KEY_SESSION_ID, currentSessionId)
+ .putString(KEY_OPENAI_MODEL, openaiModel)
+ .putString(KEY_VOLCANO_SPEECH_APP_ID, volcanoSpeechAppId)
+ .putString(KEY_VOLCANO_SPEECH_APP_TOKEN, volcanoSpeechAppToken)
.apply()
FileLogger.d("MainActivity", "密钥已安全保存到加密存储中")
@@ -227,11 +267,16 @@ class MainActivity: FlutterActivity() {
azureSpeechKey = sharedPreferences.getString(KEY_AZURE_SPEECH_KEY, "") ?: ""
azureSpeechRegion = sharedPreferences.getString(KEY_AZURE_SPEECH_REGION, "") ?: ""
volcanoAiApiKey = sharedPreferences.getString(KEY_VOLCANO_AI_API_KEY, "") ?: ""
-
+ openaiApiKey = sharedPreferences.getString(KEY_OPENAI_API_KEY, "") ?: ""
+ openaiBaseUrl = sharedPreferences.getString(KEY_OPENAI_BASE_URL, null)
+ openaiModel = sharedPreferences.getString(KEY_OPENAI_MODEL, null)
+ volcanoSpeechAppId = sharedPreferences.getString(KEY_VOLCANO_SPEECH_APP_ID, "") ?: ""
+ volcanoSpeechAppToken = sharedPreferences.getString(KEY_VOLCANO_SPEECH_APP_TOKEN, "") ?: ""
FileLogger.d("MainActivity", "已从加密存储加载密钥")
- // 检查是否成功获取所有密钥
- return azureSpeechKey.isNotEmpty() && azureSpeechRegion.isNotEmpty() && volcanoAiApiKey.isNotEmpty()
+ // 检查是否成功获取所有必要密钥
+ return azureSpeechKey.isNotEmpty() && azureSpeechRegion.isNotEmpty() &&
+ (openaiApiKey.isNotEmpty() || volcanoAiApiKey.isNotEmpty())
} catch (e: Exception) {
FileLogger.e("MainActivity", "从加密存储加载密钥失败: ${e.message}")
e.printStackTrace()
@@ -246,9 +291,10 @@ class MainActivity: FlutterActivity() {
// 初始化 FileLogger
FileLogger.init(applicationContext)
+ // 请求必要权限
+ requestRequiredPermissions()
+
// 初始化 Azure 语音服务
- azureAsrHelper = AzureAsrHelper(applicationContext)
- azureTtsHelper = AzureTtsHelper(applicationContext)
classicBluetoothHelper = ClassicBluetoothHelper(applicationContext)
// 注册广播接收器
@@ -266,7 +312,14 @@ class MainActivity: FlutterActivity() {
val voiceInteractionFilter = IntentFilter().apply {
addAction(VoiceInteractionService.ACTION_RECOGNITION_STARTED)
addAction(VoiceInteractionService.ACTION_CHAT_HISTORY_UPDATED)
+ addAction(VoiceInteractionService.ACTION_ENTER_TRANSLATION_MODE)
}
+
+ // 打印已注册的广播
+ FileLogger.d(TAG, "已注册语音交互广播：${VoiceInteractionService.ACTION_RECOGNITION_STARTED}, " +
+ "${VoiceInteractionService.ACTION_CHAT_HISTORY_UPDATED}, " +
+ "${VoiceInteractionService.ACTION_ENTER_TRANSLATION_MODE}")
+
if (android.os.Build.VERSION.SDK_INT >= android.os.Build.VERSION_CODES.UPSIDE_DOWN_CAKE) {
registerReceiver(voiceInteractionReceiver, voiceInteractionFilter, Context.RECEIVER_NOT_EXPORTED)
} else {
@@ -300,8 +353,7 @@ class MainActivity: FlutterActivity() {
setupMethodChannels(flutterEngine)
// 初始化 Azure 语音服务
- azureAsrHelper = AzureAsrHelper(this)
- azureTtsHelper = AzureTtsHelper(this)
+ classicBluetoothHelper = ClassicBluetoothHelper(this)
Log.d(TAG, "Flutter 引擎配置完成")
}
@@ -327,20 +379,6 @@ class MainActivity: FlutterActivity() {
}
)
- // Azure ASR 事件通道
- EventChannel(flutterEngine.dartExecutor.binaryMessenger, AZURE_ASR_EVENT_CHANNEL).setStreamHandler(
- object : EventChannel.StreamHandler {
- override fun onListen(arguments: Any?, events: EventChannel.EventSink?) {
- Log.d(TAG, "ASR事件通道开始监听")
- azureAsrEventSink = events
- }
-
- override fun onCancel(arguments: Any?) {
- azureAsrEventSink = null
- }
- }
- )
-
// 设置蓝牙事件通道
EventChannel(flutterEngine.dartExecutor.binaryMessenger, CLASSIC_BLUETOOTH_EVENT_CHANNEL).setStreamHandler(
object : EventChannel.StreamHandler {
@@ -367,195 +405,6 @@ class MainActivity: FlutterActivity() {
private fun setupMethodChannels(flutterEngine: FlutterEngine) {
Log.d(TAG, "开始设置方法通道")
- // 设置 Azure ASR 方法通道
- MethodChannel(flutterEngine.dartExecutor.binaryMessenger, AZURE_ASR_CHANNEL).setMethodCallHandler { call, result ->
- when (call.method) {
- "initialize" -> {
- val subscriptionKey = call.argument("subscriptionKey")
- val region = call.argument("region")
- val supportedLanguages = call.argument>("supportedLanguages")?.toTypedArray() ?: arrayOf("zh-CN", "en-US")
-
- if (subscriptionKey == null || region == null) {
- result.error("INVALID_ARGUMENTS", "订阅密钥和区域不能为空", null)
- return@setMethodCallHandler
- }
-
- try {
- val success = azureAsrHelper.initialize(subscriptionKey, region, supportedLanguages)
- result.success(success)
- } catch (e: Exception) {
- result.error("INITIALIZATION_ERROR", e.message, null)
- }
- }
- "recognizeOnce" -> {
-
- azureAsrHelper.recognizeOnce(object : AzureAsrHelper.RecognizeCallback {
- override fun onResult(text: String, detectedLanguage: String) {
- result.success(mapOf(
- "text" to text,
- "detectedLanguage" to detectedLanguage
- ))
- }
-
- override fun onError(error: String) {
- result.error("RECOGNITION_ERROR", error, null)
- }
- })
- }
- "startContinuousRecognition" -> {
- // 确保事件通道已准备好
- if (azureAsrEventSink == null) {
- result.error("EVENT_CHANNEL_NOT_READY", "事件通道未准备好，无法开始连续识别", null)
- return@setMethodCallHandler
- }
-
- val success = azureAsrHelper.startContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
- override fun onResult(text: String, detectedLanguage: String) {
- sendAsrEvent(mapOf(
- "type" to "result",
- "text" to text,
- "detectedLanguage" to detectedLanguage
- ))
- }
-
- override fun onRecognizing(recognizing: String, detectedLanguage: String) {
- sendAsrEvent(mapOf(
- "type" to "recognizing",
- "text" to recognizing,
- "detectedLanguage" to detectedLanguage
- ))
- }
-
- override fun onSessionStarted() {
- sendAsrEvent(mapOf("type" to "sessionStarted"))
- }
-
- override fun onSessionStopped() {
- sendAsrEvent(mapOf("type" to "sessionStopped"))
- }
-
- override fun onCanceled(reason: String, errorDetails: String) {
- sendAsrEvent(mapOf(
- "type" to "canceled",
- "reason" to reason,
- "errorDetails" to errorDetails
- ))
- }
-
- override fun onError(error: String) {
- sendAsrEvent(mapOf("type" to "error", "message" to error))
- }
- })
- result.success(success)
- }
- "stopContinuousRecognition" -> {
- try {
- if (!azureAsrHelper.isContinuousRecognitionActive()) {
- result.success(true)
- return@setMethodCallHandler
- }
-
- val success = azureAsrHelper.stopContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
- override fun onResult(text: String, detectedLanguage: String) {}
- override fun onRecognizing(recognizing: String, detectedLanguage: String) {}
- override fun onSessionStarted() {}
- override fun onSessionStopped() {}
- override fun onCanceled(reason: String, errorDetails: String) {}
- override fun onError(error: String) {
- result.error("STOP_ERROR", error, null)
- }
- })
- result.success(success)
- } catch (e: Exception) {
- result.error("STOP_ERROR", e.message, null)
- }
- }
- "isContinuousRecognitionActive" -> {
- result.success(azureAsrHelper.isContinuousRecognitionActive())
- }
- "dispose" -> {
- azureAsrHelper.dispose()
- result.success(true)
- }
- else -> {
- result.notImplemented()
- }
- }
- }
-
- // 设置 Azure TTS 方法通道
- MethodChannel(flutterEngine.dartExecutor.binaryMessenger, AZURE_TTS_CHANNEL).setMethodCallHandler { call, result ->
- when (call.method) {
- "initialize" -> {
- val subscriptionKey = call.argument("subscriptionKey") ?: ""
- val region = call.argument("region") ?: ""
- val language = call.argument("language") ?: "zh-CN"
-
- val success = azureTtsHelper.initialize(subscriptionKey, region, language)
- result.success(success)
- }
- "setVoice" -> {
- val voiceName = call.argument("voiceName") ?: return@setMethodCallHandler result.error("INVALID_ARGUMENTS", "语音名称不能为空", null)
- result.success(azureTtsHelper.setVoice(voiceName))
- }
- "setSpeechParams" -> {
- val rate = call.argument("rate") ?: 0
- val pitch = call.argument("pitch") ?: 0
- val volume = call.argument("volume") ?: 100
- result.success(azureTtsHelper.setSpeechParams(rate, pitch, volume))
- }
- "setAudioOutputType" -> {
- val outputTypeStr = call.argument("outputType") ?: "speaker"
- val outputType = when (outputTypeStr.lowercase()) {
- "speaker" -> AzureTtsHelper.AudioOutputType.SPEAKER
- "earpiece" -> AzureTtsHelper.AudioOutputType.EARPIECE
- "auto" -> AzureTtsHelper.AudioOutputType.AUTO
- else -> AzureTtsHelper.AudioOutputType.SPEAKER
- }
- result.success(azureTtsHelper.setAudioOutputType(outputType))
- }
- "speakText" -> {
- val text = call.argument("text") ?: return@setMethodCallHandler result.error("INVALID_ARGUMENTS", "文本不能为空", null)
-
- azureTtsHelper.speakText(text, object : AzureTtsHelper.TTSCallback {
- override fun onSuccess(message: String) {
- runOnUiThread { result.success(message) }
- }
-
- override fun onError(error: String) {
- runOnUiThread { result.error("SPEAK_ERROR", error, null) }
- }
- })
- }
- "speakSsml" -> {
- val ssml = call.argument("ssml") ?: return@setMethodCallHandler result.error("INVALID_ARGUMENTS", "SSML不能为空", null)
-
- azureTtsHelper.speakSsml(ssml, object : AzureTtsHelper.TTSCallback {
- override fun onSuccess(message: String) {
- runOnUiThread { result.success(message) }
- }
-
- override fun onError(error: String) {
- runOnUiThread { result.error("SPEAK_ERROR", error, null) }
- }
- })
- }
- "stopSpeaking" -> {
- result.success(azureTtsHelper.stopSpeaking())
- }
- "isSpeaking" -> {
- result.success(azureTtsHelper.isSpeaking())
- }
- "dispose" -> {
- azureTtsHelper.dispose()
- result.success(true)
- }
- else -> {
- result.notImplemented()
- }
- }
- }
-
// 设置语音交互方法通道
MethodChannel(flutterEngine.dartExecutor.binaryMessenger, VOICE_INTERACTION_CHANNEL).setMethodCallHandler { call, result ->
when (call.method) {
@@ -657,15 +506,6 @@ class MainActivity: FlutterActivity() {
}
}
- // ASR 事件发送方法
- private fun sendAsrEvent(event: Map) {
- if (azureAsrEventSink == null) return
-
- runOnUiThread {
- azureAsrEventSink?.success(event)
- }
- }
-
/**
* 启动语音交互服务（带配置参数）
*/
@@ -673,14 +513,14 @@ class MainActivity: FlutterActivity() {
FileLogger.d(TAG, "启动语音交互服务")
// 获取配置参数
- val key = call.argument("azure_speech_key") ?: ""
- val region = call.argument("azure_speech_region") ?: ""
- val aiKey = call.argument("volcano_ai_api_key") ?: ""
-
- // 设置 Azure Speech 配置
- azureSpeechKey = key
- azureSpeechRegion = region
- volcanoAiApiKey = aiKey
+ azureSpeechKey = call.argument("azure_speech_key") ?: ""
+ azureSpeechRegion = call.argument("azure_speech_region") ?: ""
+ openaiApiKey = call.argument("openai_api_key") ?: ""
+ openaiBaseUrl = call.argument("openai_base_url")
+ openaiModel = call.argument("openai_model")
+ volcanoSpeechAppId = call.argument("volcano_speech_app_id") ?: ""
+ volcanoSpeechAppToken = call.argument("volcano_speech_app_token") ?: ""
+
// 保存密钥到安全存储
saveKeysToSecureStorage(applicationContext)
@@ -755,14 +595,19 @@ class MainActivity: FlutterActivity() {
private fun sendVoiceInteractionEvent(event: Map) {
if (voiceInteractionEventSink == null) {
Log.e(TAG, "无法发送语音交互事件：事件通道未准备好")
+ FileLogger.e(TAG, "无法发送语音交互事件：事件通道未准备好")
return
}
+ FileLogger.d(TAG, "准备发送事件到Flutter: ${event["type"]}")
+
runOnUiThread {
try {
voiceInteractionEventSink?.success(event)
+ FileLogger.d(TAG, "成功发送事件到Flutter: ${event["type"]}")
} catch (e: Exception) {
- Log.e(TAG, "发送语音交互事件到Flutter失败: ${e.message}", e)
+ Log.e(TAG, "发送语音交互事件到Flutter失败: ${e.message}")
+ FileLogger.e(TAG, "发送语音交互事件到Flutter失败: ${e.message}")
}
}
}
@@ -799,11 +644,58 @@ class MainActivity: FlutterActivity() {
}
// 释放资源
- azureAsrHelper.dispose()
- azureTtsHelper.dispose()
classicBluetoothHelper.dispose()
super.onDestroy()
}
+
+ /**
+ * 请求必要权限
+ */
+ private fun requestRequiredPermissions() {
+ val permissionsToRequest = ArrayList()
+
+ for (permission in REQUIRED_PERMISSIONS) {
+ if (ContextCompat.checkSelfPermission(this, permission) != PackageManager.PERMISSION_GRANTED) {
+ permissionsToRequest.add(permission)
+ }
+ }
+
+ if (permissionsToRequest.isNotEmpty()) {
+ ActivityCompat.requestPermissions(
+ this,
+ permissionsToRequest.toTypedArray(),
+ PERMISSION_REQUEST_CODE
+ )
+ }
+ }
+
+ /**
+ * 处理权限请求结果
+ */
+ override fun onRequestPermissionsResult(
+ requestCode: Int,
+ permissions: Array,
+ grantResults: IntArray
+ ) {
+ super.onRequestPermissionsResult(requestCode, permissions, grantResults)
+
+ if (requestCode == PERMISSION_REQUEST_CODE) {
+ val deniedPermissions = ArrayList()
+
+ for (i in permissions.indices) {
+ if (grantResults[i] != PackageManager.PERMISSION_GRANTED) {
+ deniedPermissions.add(permissions[i])
+ }
+ }
+
+ if (deniedPermissions.isNotEmpty()) {
+ // 记录未授权的权限
+ FileLogger.w(TAG, "未授权的权限: ${deniedPermissions.joinToString()}")
+ } else {
+ FileLogger.d(TAG, "所有必要权限已授权")
+ }
+ }
+ }
}
diff --git a/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionHandler copy.kt.bak b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionHandler copy.kt.bak
new file mode 100644
index 000000000..1a9e2df13
--- /dev/null
+++ b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionHandler copy.kt.bak
@@ -0,0 +1,456 @@
+package com.yunqiinnovation.deepsound
+
+import android.content.Context
+import org.json.JSONArray
+import org.json.JSONObject
+import android.util.Log
+import com.yunqiinnovation.deepsound.core.utils.FileLogger
+import com.yunqiinnovation.azure_speech.AzureAsrHelper
+import com.yunqiinnovation.volcano_speech.VolcanoTtsHelper
+import com.yunqiinnovation.open_ai_service.OpenAIService
+
+
+/**
+ * 语音交互处理器 - 处理语音识别、TTS和AI对话相关逻辑
+ */
+class VoiceInteractionHandler(
+ private val context: Context,
+ private val azureSpeechKey: String,
+ private val azureSpeechRegion: String,
+ private val openaiApiKey: String,
+ private val openaiBaseUrl: String = "",
+ private val openaiModel: String = "",
+ private val volcanoSpeechAppId: String,
+ private val volcanoSpeechAppToken: String
+) {
+ private val TAG = "VoiceInteractionHandler"
+
+ // Azure服务
+ private var azureAsrHelper: AzureAsrHelper? = null
+ private var volcanoTtsHelper: VolcanoTtsHelper? = null
+
+ // OpenAI服务
+ private val openAIService = OpenAIService()
+
+
+ // 语音功能处理
+ private val voiceFunctionHandler = VoiceFunctionHandler(openAIService, context)
+
+ // 当前用户输入
+ private var currentUserInput = ""
+
+ // 状态
+ private var isInitialized = false
+ var isRecognitionActive = false
+ private set
+ var isTtsSpeaking = false
+ private set
+ var hasSpeechDetected = false
+ private set
+
+ // 回调
+ private var callback: InteractionCallback? = null
+
+ /**
+ * 初始化
+ */
+ fun initialize(): Boolean {
+ if (isInitialized) return true
+
+ try {
+ // 初始化Azure ASR
+ azureAsrHelper = AzureAsrHelper(context).apply {
+ initialize(azureSpeechKey, azureSpeechRegion)
+ }
+
+ // 初始化Volcano TTS (使用大模型TTS)
+ volcanoTtsHelper = VolcanoTtsHelper(context).apply {
+ // 这里需要替换为实际的Volcano SDK初始化参数
+ // 暂时使用假参数，实际使用时需要替换为真实值
+ initialize(volcanoSpeechAppId, volcanoSpeechAppToken, "volc.bigasr.sauc.duration")
+
+ }
+
+ // 初始化OpenAI服务
+ openAIService.initialize(openaiApiKey, openaiBaseUrl, openaiModel)
+
+ // 初始化语音功能处理器
+ voiceFunctionHandler.initialize()
+
+ isInitialized = true
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "初始化失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置回调
+ */
+ fun setCallback(callback: InteractionCallback) {
+ this.callback = callback
+ }
+
+ /**
+ * 开始语音识别
+ */
+ fun startRecognition() {
+ if (isRecognitionActive) return
+
+ // 检查录音权限
+ if (!checkRecordAudioPermission()) {
+ callback?.onError("需要录音权限，请在设置中授予权限")
+ return
+ }
+
+ isRecognitionActive = true
+ hasSpeechDetected = false
+ notifyStateChanged()
+
+ try {
+ azureAsrHelper?.startContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
+ override fun onRecognizing(recognizing: String, detectedLanguage: String) {
+ if (recognizing.isNotEmpty()) {
+ hasSpeechDetected = true
+ stopTts()
+ notifyStateChanged()
+ }
+ }
+
+ override fun onResult(result: String, detectedLanguage: String) {
+ if (result.isNotEmpty()) {
+ notifyStateChanged()
+
+ processWithOpenAI(result)
+
+ }
+
+ // 重置状态，继续识别
+ hasSpeechDetected = false
+ }
+
+ override fun onSessionStarted() {
+ notifyStateChanged()
+ }
+
+ override fun onSessionStopped() {
+ isRecognitionActive = false
+ notifyStateChanged()
+ }
+
+ override fun onCanceled(reason: String, errorDetails: String) {
+ isRecognitionActive = false
+ notifyStateChanged()
+ }
+
+ override fun onError(error: String) {
+ isRecognitionActive = false
+ callback?.onError("语音识别出错")
+ notifyStateChanged()
+ }
+
+ override fun onSuccess(message: String) {
+ // 处理成功事件
+ }
+ })
+ } catch (e: Exception) {
+ isRecognitionActive = false
+ FileLogger.e(TAG, "启动语音识别失败: ${e.message}", e)
+ callback?.onError("启动语音识别失败")
+ notifyStateChanged()
+ }
+ }
+
+ /**
+ * 停止语音识别
+ */
+ fun stopRecognition() {
+ if (!isRecognitionActive) return
+
+ FileLogger.d(TAG, "停止语音识别")
+
+ try {
+ azureAsrHelper?.stopContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
+ override fun onResult(result: String, detectedLanguage: String) {}
+ override fun onRecognizing(recognizing: String, detectedLanguage: String) {}
+ override fun onSessionStarted() {}
+ override fun onSessionStopped() {
+ isRecognitionActive = false
+ FileLogger.d(TAG, "语音识别会话已停止")
+ notifyStateChanged()
+ }
+ override fun onCanceled(reason: String, errorDetails: String) {
+ isRecognitionActive = false
+ FileLogger.d(TAG, "语音识别已取消: $reason")
+ notifyStateChanged()
+ }
+ override fun onError(error: String) {
+ isRecognitionActive = false
+ FileLogger.e(TAG, "停止语音识别时出错: $error")
+ notifyStateChanged()
+ }
+ override fun onSuccess(message: String) {
+ isRecognitionActive = false
+ FileLogger.d(TAG, "语音识别已停止: $message")
+ notifyStateChanged()
+ }
+ })
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "停止语音识别异常: ${e.message}", e)
+ // 确保状态一致性
+ isRecognitionActive = false
+ notifyStateChanged()
+ }
+ }
+
+ /**
+ * 使用OpenAI处理语音识别结果
+ */
+ private fun processWithOpenAI(text: String) {
+ // 保存当前用户输入，用于后续同步聊天记录
+ currentUserInput = text
+
+ Thread {
+ try {
+ val messages = JSONArray().apply {
+ put(openAIService.createUserMessage(text))
+ }
+
+ // 创建响应构建器
+ val responseBuilder = StringBuilder()
+
+ openAIService.sendMessageStream(
+ messages = messages,
+ callback = object : OpenAIService.StreamCallback {
+ override fun onToken(token: String) {
+ // 累加响应内容
+ responseBuilder.append(token)
+ }
+
+ override fun onComplete() {
+ // 处理完整响应
+ val response = responseBuilder.toString()
+ if (response.isNotEmpty()) {
+ // 播放AI回复
+ Log.d(TAG, "AI 回复: $response")
+
+ speakAIResponse(response)
+
+ // 同步聊天记录到Flutter端
+ sendChatHistoryUpdate("personal_assistant", text, response)
+ }
+ }
+
+ override fun onError(e: Exception) {
+ FileLogger.e(TAG, "AI流式处理出错: ${e.message}", e)
+ callback?.onError("AI处理出错")
+ }
+
+ override fun onFunctionCall(call: JSONObject) {
+ FileLogger.d(TAG, "收到函数调用请求: ${call.getString("name")}")
+
+ // 使用函数处理器处理函数调用
+ val handled = voiceFunctionHandler.handleFunctionCall(
+ functionCall = call,
+ messages = messages,
+ callback = object : VoiceFunctionHandler.FunctionCallCallback {
+ override fun onTokenReceived(token: String) {
+ responseBuilder.append(token)
+ }
+
+ override fun onComplete() {
+ val response = responseBuilder.toString()
+ if (response.isNotEmpty()) {
+ // 播放AI回复
+ Log.d(TAG, "AI Function Call 回复: $response")
+ speakAIResponse(response)
+
+ // 同步聊天记录到Flutter端
+ sendChatHistoryUpdate("personal_assistant", text, response)
+ }
+ notifyStateChanged()
+ }
+
+ override fun onError(message: String) {
+ FileLogger.e(TAG, "函数处理出错: $message")
+ callback?.onError(message)
+ }
+
+ override fun onFunctionCall(nestedCall: JSONObject) {
+ FileLogger.d(TAG, "收到嵌套函数调用: ${nestedCall.getString("name")}")
+ // 不处理嵌套函数调用，直接返回错误提示
+ // speakAIResponse("抱歉，暂不支持嵌套函数调用")
+ }
+
+ override fun onExitWithMessage(farewell: String) {
+ // 播放退出消息
+ speakAIResponse(farewell)
+
+ // 同步聊天记录
+ sendChatHistoryUpdate("personal_assistant", text, farewell)
+
+ // 停止语音识别
+ stopRecognition()
+ }
+ }
+ )
+
+ if (!handled) {
+ // 如果函数没有被处理，作为普通文本处理
+ FileLogger.d(TAG, "函数未处理，作为普通文本处理")
+ speakAIResponse("我无法处理这个请求")
+ sendChatHistoryUpdate("personal_assistant", text, "我无法处理这个请求")
+ }
+ }
+ }
+ )
+
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "AI处理出错: ${e.message}", e)
+ callback?.onError("AI处理出错")
+ }
+ }.start()
+ }
+
+ /**
+ * 播放TTS
+ */
+ fun playTts(text: String, callback: TtsCallback? = null) {
+ isTtsSpeaking = true
+ notifyStateChanged()
+
+ volcanoTtsHelper?.speak(text, object : VolcanoTtsHelper.TTSCallback {
+ override fun onStart(reqId: String) {
+ // TTS开始播放事件
+ }
+
+ override fun onProgress(reqId: String, progress: Double) {
+ // TTS播放进度事件
+ }
+
+ override fun onComplete(reqId: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ callback?.onComplete()
+ }
+
+ override fun onError(reqId: String, errorCode: Int, errorMsg: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ callback?.onError(errorMsg)
+ }
+ })
+ }
+
+ /**
+ * 停止TTS播放
+ */
+ fun stopTts() {
+ if (isTtsSpeaking) {
+ volcanoTtsHelper?.stop()
+ isTtsSpeaking = false
+ notifyStateChanged()
+ }
+ }
+
+ /**
+ * 播放AI回复
+ */
+ private fun speakAIResponse(text: String) {
+ isTtsSpeaking = true
+ notifyStateChanged()
+
+ volcanoTtsHelper?.speak(text, object : VolcanoTtsHelper.TTSCallback {
+ override fun onStart(reqId: String) {
+ // TTS开始播放事件
+ }
+
+ override fun onProgress(reqId: String, progress: Double) {
+ // TTS播放进度事件
+ }
+
+ override fun onComplete(reqId: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ }
+
+ override fun onError(reqId: String, errorCode: Int, errorMsg: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ }
+ })
+ }
+
+ /**
+ * 释放资源
+ */
+ fun dispose() {
+ // 停止语音识别
+ stopRecognition()
+
+ // 停止TTS播放
+ stopTts()
+
+ // 释放Azure资源
+ azureAsrHelper?.let {
+ FileLogger.d(TAG, "关闭Azure ASR服务")
+ it.dispose()
+ }
+
+ volcanoTtsHelper?.let {
+ FileLogger.d(TAG, "关闭Volcano TTS服务")
+ it.release()
+ }
+
+ FileLogger.d(TAG, "语音交互处理器资源已释放")
+ }
+
+ /**
+ * 检查录音权限
+ */
+ private fun checkRecordAudioPermission(): Boolean {
+ val permission = android.Manifest.permission.RECORD_AUDIO
+ val result = context.checkCallingOrSelfPermission(permission)
+ return result == android.content.pm.PackageManager.PERMISSION_GRANTED
+ }
+
+ /**
+ * 通知状态变化
+ */
+ private fun notifyStateChanged() {
+ callback?.onStateChanged(isRecognitionActive, isTtsSpeaking, hasSpeechDetected)
+ }
+
+ /**
+ * 发送聊天历史更新
+ */
+ private fun sendChatHistoryUpdate(agentId: String, userMessage: String, assistantMessage: String) {
+ val intent = android.content.Intent(VoiceInteractionService.ACTION_CHAT_HISTORY_UPDATED).apply {
+ putExtra("agentId", agentId)
+ putExtra("userMessage", userMessage)
+ putExtra("assistantMessage", assistantMessage)
+ putExtra("timestamp", System.currentTimeMillis())
+ }
+
+ // 发送广播
+ context.sendBroadcast(intent)
+ }
+
+ /**
+ * 交互回调接口
+ */
+ interface InteractionCallback {
+ fun onStateChanged(isRecognitionActive: Boolean, isTtsSpeaking: Boolean, hasSpeechDetected: Boolean)
+ fun onError(message: String)
+ fun onPromptRequest(message: String)
+ }
+
+ /**
+ * TTS回调接口
+ */
+ interface TtsCallback {
+ fun onComplete()
+ fun onError(error: String)
+ }
+}
\ No newline at end of file
diff --git a/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionHandler.kt b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionHandler.kt
new file mode 100644
index 000000000..b885117c0
--- /dev/null
+++ b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionHandler.kt
@@ -0,0 +1,416 @@
+package com.yunqiinnovation.deepsound
+
+import android.content.BroadcastReceiver
+import android.content.Context
+import android.content.Intent
+import android.content.IntentFilter
+import org.json.JSONArray
+import org.json.JSONObject
+import android.util.Log
+import com.yunqiinnovation.deepsound.core.utils.FileLogger
+import com.yunqiinnovation.azure_speech.AzureAsrHelper
+import com.yunqiinnovation.azure_speech.AzureTtsHelper
+import com.yunqiinnovation.open_ai_service.OpenAIService
+import com.yunqiinnovation.open_ai_service.SystemFunctionHandler
+
+
+/**
+ * 语音交互处理器 - 处理语音识别、TTS和AI对话相关逻辑
+ */
+class VoiceInteractionHandler(
+ private val context: Context,
+ private val azureSpeechKey: String,
+ private val azureSpeechRegion: String,
+ private val openaiApiKey: String,
+ private val openaiBaseUrl: String = "",
+ private val openaiModel: String = "",
+ private val volcanoSpeechAppId: String,
+ private val volcanoSpeechAppToken: String
+) {
+ private val TAG = "VoiceInteractionHandler"
+
+ // Azure服务
+ private var azureAsrHelper: AzureAsrHelper? = null
+ private var azureTtsHelper: AzureTtsHelper? = null
+
+ // OpenAI服务
+ private val openAIService = OpenAIService(context.applicationContext)
+
+ // 当前用户输入
+ private var currentUserInput = ""
+
+ // 状态
+ private var isInitialized = false
+ var isRecognitionActive = false
+ private set
+ var isTtsSpeaking = false
+ private set
+ var hasSpeechDetected = false
+ private set
+
+ // 回调
+ private var callback: InteractionCallback? = null
+
+ // 广播接收器
+ private val exitInteractionReceiver = object : BroadcastReceiver() {
+ override fun onReceive(context: Context, intent: Intent) {
+ if (intent.action == SystemFunctionHandler.ACTION_EXIT_INTERACTION) {
+ Log.d(TAG, "收到退出交互广播")
+ stopRecognition()
+
+ }
+ }
+ }
+
+ /**
+ * 初始化
+ */
+ fun initialize(): Boolean {
+ if (isInitialized) return true
+
+ try {
+ // 初始化Azure ASR
+ azureAsrHelper = AzureAsrHelper(context).apply {
+ initialize(azureSpeechKey, azureSpeechRegion)
+ }
+
+ // 初始化Azure TTS
+ azureTtsHelper = AzureTtsHelper(context).apply {
+ initialize(azureSpeechKey, azureSpeechRegion)
+ }
+
+ // 初始化OpenAI服务
+ openAIService.initialize(openaiApiKey, openaiBaseUrl, openaiModel)
+
+ // 注册广播接收器
+ try {
+ Log.d(TAG, "注册退出交互广播接收器，包名=${context.packageName}, action=${SystemFunctionHandler.ACTION_EXIT_INTERACTION}")
+ context.registerReceiver(
+ exitInteractionReceiver,
+ IntentFilter(SystemFunctionHandler.ACTION_EXIT_INTERACTION),
+ Context.RECEIVER_NOT_EXPORTED
+ )
+ Log.d(TAG, "退出交互广播接收器注册成功")
+ } catch (e: Exception) {
+ // 广播注册失败不应该影响整个应用初始化
+ Log.e(TAG, "注册退出交互广播接收器失败: ${e.message}", e)
+ }
+
+ isInitialized = true
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "初始化失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置回调
+ */
+ fun setCallback(callback: InteractionCallback) {
+ this.callback = callback
+ }
+
+ /**
+ * 开始语音识别
+ */
+ fun startRecognition() {
+ if (isRecognitionActive) return
+
+ // 检查录音权限
+ if (!checkRecordAudioPermission()) {
+ callback?.onError("需要录音权限，请在设置中授予权限")
+ return
+ }
+
+ isRecognitionActive = true
+ hasSpeechDetected = false
+ notifyStateChanged()
+
+ try {
+ azureAsrHelper?.startContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
+ override fun onRecognizing(recognizing: String, detectedLanguage: String) {
+ if (recognizing.isNotEmpty()) {
+ hasSpeechDetected = true
+ stopTts()
+ notifyStateChanged()
+ }
+ }
+
+ override fun onResult(result: String, detectedLanguage: String) {
+ if (result.isNotEmpty()) {
+ notifyStateChanged()
+
+ processWithOpenAI(result)
+
+ }
+
+ // 重置状态，继续识别
+ hasSpeechDetected = false
+ }
+
+ override fun onSessionStarted() {
+ notifyStateChanged()
+ }
+
+ override fun onSessionStopped() {
+ isRecognitionActive = false
+ notifyStateChanged()
+ }
+
+ override fun onCanceled(reason: String, errorDetails: String) {
+ isRecognitionActive = false
+ notifyStateChanged()
+ }
+
+ override fun onError(error: String) {
+ isRecognitionActive = false
+ callback?.onError("语音识别出错")
+ notifyStateChanged()
+ }
+
+ override fun onSuccess(message: String) {
+ // 处理成功事件
+ }
+ })
+ } catch (e: Exception) {
+ isRecognitionActive = false
+ FileLogger.e(TAG, "启动语音识别失败: ${e.message}", e)
+ callback?.onError("启动语音识别失败")
+ notifyStateChanged()
+ }
+ }
+
+ /**
+ * 停止语音识别
+ */
+ fun stopRecognition() {
+ if (!isRecognitionActive) return
+
+ FileLogger.d(TAG, "停止语音识别")
+
+ try {
+ azureAsrHelper?.stopContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
+ override fun onResult(result: String, detectedLanguage: String) {}
+ override fun onRecognizing(recognizing: String, detectedLanguage: String) {}
+ override fun onSessionStarted() {}
+ override fun onSessionStopped() {
+ isRecognitionActive = false
+ FileLogger.d(TAG, "语音识别会话已停止")
+ notifyStateChanged()
+ }
+ override fun onCanceled(reason: String, errorDetails: String) {
+ isRecognitionActive = false
+ FileLogger.d(TAG, "语音识别已取消: $reason")
+ notifyStateChanged()
+ }
+ override fun onError(error: String) {
+ isRecognitionActive = false
+ FileLogger.e(TAG, "停止语音识别时出错: $error")
+ notifyStateChanged()
+ }
+ override fun onSuccess(message: String) {
+ isRecognitionActive = false
+ FileLogger.d(TAG, "语音识别已停止: $message")
+ notifyStateChanged()
+ }
+ })
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "停止语音识别异常: ${e.message}", e)
+ // 确保状态一致性
+ isRecognitionActive = false
+ notifyStateChanged()
+ }
+ }
+
+ /**
+ * 使用OpenAI处理语音识别结果
+ */
+ private fun processWithOpenAI(text: String) {
+ // 保存当前用户输入，用于后续同步聊天记录
+ currentUserInput = text
+
+ Thread {
+ try {
+ val messages = JSONArray().apply {
+ put(openAIService.createUserMessage(text))
+ }
+
+ // 创建响应构建器
+ val responseBuilder = StringBuilder()
+
+ openAIService.sendMessageStream(
+ messages = messages,
+ callback = object : OpenAIService.StreamCallback {
+ override fun onToken(token: String) {
+ // 累加响应内容
+ responseBuilder.append(token)
+ }
+
+ override fun onComplete() {
+ // 处理完整响应
+ val response = responseBuilder.toString()
+ if (response.isNotEmpty()) {
+ // 播放AI回复
+ Log.d(TAG, "AI 回复: $response")
+
+ speakAIResponse(response)
+
+ // 同步聊天记录到Flutter端
+ sendChatHistoryUpdate("personal_assistant", text, response)
+ }
+ }
+
+ override fun onError(e: Exception) {
+ FileLogger.e(TAG, "AI流式处理出错: ${e.message}", e)
+ callback?.onError("AI处理出错")
+ }
+
+ override fun onFunctionCall(call: JSONObject) {
+ FileLogger.d(TAG, "processWithOpenAI 收到函数调用请求: ${call.getString("name")}")
+
+
+ }
+ }
+ )
+
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "AI处理出错: ${e.message}", e)
+ callback?.onError("AI处理出错")
+ }
+ }.start()
+ }
+
+ /**
+ * 播放TTS
+ */
+ fun playTts(text: String, callback: TtsCallback? = null) {
+ isTtsSpeaking = true
+ notifyStateChanged()
+
+ azureTtsHelper?.speakText(text, object : AzureTtsHelper.TTSCallback {
+ override fun onSuccess(message: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ callback?.onComplete()
+ }
+
+ override fun onError(error: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ callback?.onError(error)
+ }
+ })
+ }
+
+ /**
+ * 停止TTS播放
+ */
+ fun stopTts() {
+ if (isTtsSpeaking) {
+ azureTtsHelper?.stopSpeaking()
+ isTtsSpeaking = false
+ notifyStateChanged()
+ }
+ }
+
+ /**
+ * 播放AI回复
+ */
+ private fun speakAIResponse(text: String) {
+ isTtsSpeaking = true
+ notifyStateChanged()
+
+ azureTtsHelper?.speakText(text, object : AzureTtsHelper.TTSCallback {
+ override fun onSuccess(message: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ }
+
+ override fun onError(error: String) {
+ isTtsSpeaking = false
+ notifyStateChanged()
+ }
+ })
+ }
+
+ /**
+ * 释放资源
+ */
+ fun dispose() {
+ // 停止语音识别
+ stopRecognition()
+
+ // 停止TTS播放
+ stopTts()
+
+ // 注销广播接收器
+ try {
+ context.unregisterReceiver(exitInteractionReceiver)
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "注销广播接收器失败: ${e.message}", e)
+ }
+
+ // 释放Azure资源
+ azureAsrHelper?.let {
+ FileLogger.d(TAG, "关闭Azure ASR服务")
+ it.dispose()
+ }
+
+ azureTtsHelper?.let {
+ FileLogger.d(TAG, "关闭Azure TTS服务")
+ it.dispose()
+ }
+
+ FileLogger.d(TAG, "语音交互处理器资源已释放")
+ }
+
+ /**
+ * 检查录音权限
+ */
+ private fun checkRecordAudioPermission(): Boolean {
+ val permission = android.Manifest.permission.RECORD_AUDIO
+ val result = context.checkCallingOrSelfPermission(permission)
+ return result == android.content.pm.PackageManager.PERMISSION_GRANTED
+ }
+
+ /**
+ * 通知状态变化
+ */
+ private fun notifyStateChanged() {
+ callback?.onStateChanged(isRecognitionActive, isTtsSpeaking, hasSpeechDetected)
+ }
+
+ /**
+ * 发送聊天历史更新
+ */
+ private fun sendChatHistoryUpdate(agentId: String, userMessage: String, assistantMessage: String) {
+ val intent = android.content.Intent(VoiceInteractionService.ACTION_CHAT_HISTORY_UPDATED).apply {
+ putExtra("agentId", agentId)
+ putExtra("userMessage", userMessage)
+ putExtra("assistantMessage", assistantMessage)
+ putExtra("timestamp", System.currentTimeMillis())
+ setPackage(context.packageName)
+ }
+
+ // 发送广播
+ context.sendBroadcast(intent)
+ }
+
+ /**
+ * 交互回调接口
+ */
+ interface InteractionCallback {
+ fun onStateChanged(isRecognitionActive: Boolean, isTtsSpeaking: Boolean, hasSpeechDetected: Boolean)
+ fun onError(message: String)
+ fun onPromptRequest(message: String)
+ }
+
+ /**
+ * TTS回调接口
+ */
+ interface TtsCallback {
+ fun onComplete()
+ fun onError(error: String)
+ }
+}
\ No newline at end of file
diff --git a/android/app/src/main/kotlin/com/example/deep_voice/VoiceInteractionService.kt b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionService.kt
similarity index 67%
rename from android/app/src/main/kotlin/com/example/deep_voice/VoiceInteractionService.kt
rename to android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionService.kt
index f469c136e..dbe64c79e 100644
--- a/android/app/src/main/kotlin/com/example/deep_voice/VoiceInteractionService.kt
+++ b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/VoiceInteractionService.kt
@@ -22,16 +22,21 @@ import android.os.Handler
import android.os.Looper
import java.util.concurrent.atomic.AtomicBoolean
import org.json.JSONArray
+import org.json.JSONObject
import android.media.MediaPlayer
import android.media.AudioAttributes
import android.net.Uri
import com.yunqiinnovation.deepsound.core.utils.FileLogger
+import com.yunqiinnovation.azure_speech.AzureAsrHelper
+import com.yunqiinnovation.volcano_speech.VolcanoTtsHelper
+import com.yunqiinnovation.open_ai_service.OpenAIService
+
/**
* 后台语音交互 Service：
* 1) 前台服务，确保不会被系统轻易杀死
* 2) MediaSession 捕获蓝牙耳机按键
- * 3) 处理录音/语音识别
+ * 3) 负责唤醒控制和服务生命周期管理
*/
class VoiceInteractionService : Service() {
@@ -41,7 +46,7 @@ class VoiceInteractionService : Service() {
private const val CHANNEL_ID = "voice_interaction_channel"
// 语音识别超时时间（毫秒）
- private const val RECOGNITION_TIMEOUT = 8000L
+ private const val RECOGNITION_TIMEOUT = 10000L
// 用于跟踪服务是否正在运行
private val isRunning = AtomicBoolean(false)
@@ -53,14 +58,12 @@ class VoiceInteractionService : Service() {
const val ACTION_RECOGNITION_STARTED = "com.yunqiinnovation.deepsound.ACTION_RECOGNITION_STARTED"
const val ACTION_PAUSE_VOICE_INTERACTION = "com.yunqiinnovation.deepsound.ACTION_PAUSE_VOICE_INTERACTION"
const val ACTION_CHAT_HISTORY_UPDATED = "com.yunqiinnovation.deepsound.ACTION_CHAT_HISTORY_UPDATED"
+ const val ACTION_ENTER_TRANSLATION_MODE = "com.yunqiinnovation.deepsound.ACTION_ENTER_TRANSLATION_MODE"
}
// 服务状态
private var isActive = false // 服务是否活跃
- private var isRecognitionActive = false // 语音识别是否活跃
private var isTimeoutPaused = false // 是否因超时暂停
- private var hasSpeechDetected = false // 是否检测到语音
- private var isTtsSpeaking = false // 是否正在播放TTS
// 按键处理
private var lastKeyEventTime = 0L
@@ -69,16 +72,12 @@ class VoiceInteractionService : Service() {
// 活动时间
private var lastActivityTime = 0L
- // 当前用户输入
- private var currentUserInput = ""
-
-
// 服务组件
private lateinit var mediaSession: MediaSessionCompat
private lateinit var audioManager: AudioManager
- private lateinit var azureAsrHelper: AzureAsrHelper
- private lateinit var azureTtsHelper: AzureTtsHelper
- private lateinit var volcanoAIService: VolcanoAIService
+
+ // 语音交互处理器
+ private lateinit var voiceInteractionHandler: VoiceInteractionHandler
// 定时器
private val handler = Handler(Looper.getMainLooper())
@@ -88,15 +87,6 @@ class VoiceInteractionService : Service() {
handler.postDelayed(this, 1000) // 每秒执行一次
}
}
-
- // 系统提示词
- private val systemPrompt = """
- 你是一个智能语音助手，能够简洁明了地回答用户的问题。
- 请保持回答简短、准确，避免过长的解释。
- 如果用户的问题不清楚，请礼貌地请求澄清。
- 不要使用复杂的术语，除非用户明确要求。
- 用户用语音和你交互.
- """.trimIndent()
// 添加媒体播放器
private var audioPlayer: MiniMediaPlayer? = null
@@ -115,7 +105,9 @@ class VoiceInteractionService : Service() {
audioManager = getSystemService(Context.AUDIO_SERVICE) as AudioManager
FileLogger.d(TAG, "AudioManager初始化完成")
- initServices()
+ // 初始化语音交互处理器
+ initVoiceInteractionHandler()
+
initMediaSession()
registerMediaButtonReceiver()
@@ -125,7 +117,6 @@ class VoiceInteractionService : Service() {
// 设置为媒体播放状态
setPlaybackState(PlaybackStateCompat.STATE_PAUSED)
- // FileLogger.d(TAG, "设置播放状态为STATE_PAUSED")
// 启动监控和前台服务
startMonitoring()
@@ -139,28 +130,26 @@ class VoiceInteractionService : Service() {
*/
private fun resetState() {
isActive = false
- isRecognitionActive = false
isTimeoutPaused = false
- hasSpeechDetected = false
- isTtsSpeaking = false
}
/**
- * 初始化所有服务
+ * 初始化语音交互处理器
*/
- private fun initServices() {
- // 创建新的Azure服务实例
- FileLogger.d(TAG, "创建新的Azure服务实例")
- azureAsrHelper = AzureAsrHelper(this)
- azureTtsHelper = AzureTtsHelper(this)
+ private fun initVoiceInteractionHandler() {
+ FileLogger.d(TAG, "初始化语音交互处理器")
// 尝试从静态变量获取配置
var subscriptionKey = MainActivity.azureSpeechKey
var serviceRegion = MainActivity.azureSpeechRegion
- var volcanoKey = MainActivity.volcanoAiApiKey
+ var openaiKey = MainActivity.openaiApiKey
+ var openaiBaseUrl = MainActivity.openaiBaseUrl ?: "" // OpenAI API基本URL
+ var openaiModel = MainActivity.openaiModel ?: "" // OpenAI模型
+ var volcanoSpeechAppId = MainActivity.volcanoSpeechAppId ?: ""
+ var volcanoSpeechAppToken = MainActivity.volcanoSpeechAppToken ?: ""
// 如果静态变量中没有配置，尝试从加密存储中加载
- if (subscriptionKey.isEmpty() || serviceRegion.isEmpty() || volcanoKey.isEmpty()) {
+ if (subscriptionKey.isEmpty() || serviceRegion.isEmpty() || openaiKey.isEmpty()) {
FileLogger.d(TAG, "静态变量中的配置信息不完整，尝试从加密存储加载")
// 从加密存储加载密钥
@@ -170,7 +159,11 @@ class VoiceInteractionService : Service() {
// 更新本地变量
subscriptionKey = MainActivity.azureSpeechKey
serviceRegion = MainActivity.azureSpeechRegion
- volcanoKey = MainActivity.volcanoAiApiKey
+ openaiKey = MainActivity.openaiApiKey
+ openaiBaseUrl = MainActivity.openaiBaseUrl ?: ""
+ openaiModel = MainActivity.openaiModel ?: ""
+ volcanoSpeechAppId = MainActivity.volcanoSpeechAppId ?: ""
+ volcanoSpeechAppToken = MainActivity.volcanoSpeechAppToken ?: ""
FileLogger.d(TAG, "已从加密存储加载配置信息")
} else {
@@ -178,28 +171,34 @@ class VoiceInteractionService : Service() {
}
}
- // 初始化语音服务
- if (subscriptionKey.isNotEmpty() && serviceRegion.isNotEmpty()) {
- // 初始化ASR
- azureAsrHelper.initialize(subscriptionKey, serviceRegion, arrayOf("zh-CN"))
+ // 初始化语音交互处理器
+ voiceInteractionHandler = VoiceInteractionHandler(applicationContext,
+ subscriptionKey, serviceRegion,
+ openaiKey, openaiBaseUrl, openaiModel,
+ volcanoSpeechAppId, volcanoSpeechAppToken)
+
+ // 初始化回调
+ voiceInteractionHandler.setCallback(object : VoiceInteractionHandler.InteractionCallback {
+ override fun onStateChanged(isRecognitionActive: Boolean, isTtsSpeaking: Boolean, hasSpeechDetected: Boolean) {
+ // 更新活动时间
+ updateLastActivityTime()
+ }
- // 初始化TTS
- azureTtsHelper.initialize(subscriptionKey, serviceRegion, "zh-CN")
+ override fun onError(message: String) {
+ playNotification(message)
+ }
- FileLogger.d(TAG, "Azure语音服务已初始化")
- } else {
- FileLogger.e(TAG, "Azure配置信息不完整，无法初始化Azure服务")
- }
-
- // 初始化火山AI服务
- volcanoAIService = VolcanoAIService()
+ override fun onPromptRequest(message: String) {
+ playPrompt(message)
+ }
+ })
- // 初始化火山AI服务
- if (volcanoKey.isNotEmpty()) {
- volcanoAIService.initialize(volcanoKey)
- FileLogger.d(TAG, "火山AI服务已初始化")
+ // 初始化处理器
+ val initialized = voiceInteractionHandler.initialize()
+ if (initialized) {
+ FileLogger.d(TAG, "语音交互处理器初始化成功")
} else {
- FileLogger.e(TAG, "火山AI配置信息不完整，无法初始化火山AI服务")
+ FileLogger.e(TAG, "语音交互处理器初始化失败")
}
}
@@ -302,24 +301,23 @@ class VoiceInteractionService : Service() {
if (!isActive) {
isActive = true
}
-
+ // FileLogger.d(TAG, "服务状态: isActive=${isActive}, isRecognitionActive=${voiceInteractionHandler.isRecognitionActive}, isTimeoutPaused=${isTimeoutPaused}")
// 检查语音识别状态
- if (isRecognitionActive) {
+ if (voiceInteractionHandler.isRecognitionActive) {
val currentTime = System.currentTimeMillis()
val elapsedTime = currentTime - lastActivityTime
-
- // 如果超过5秒没有检测到语音，且不在TTS播放中，暂停语音识别
- if (!hasSpeechDetected && !isTtsSpeaking && elapsedTime >= RECOGNITION_TIMEOUT) {
+ // FileLogger.d(TAG, "hasSpeechDetected=${voiceInteractionHandler.hasSpeechDetected}, isTtsSpeaking=${voiceInteractionHandler.isTtsSpeaking}, elapsedTime=${elapsedTime}")
+ // 如果超过指定时间没有检测到语音，且不在TTS播放中，暂停语音识别
+ if (!voiceInteractionHandler.hasSpeechDetected &&
+ !voiceInteractionHandler.isTtsSpeaking &&
+ elapsedTime >= RECOGNITION_TIMEOUT) {
+
FileLogger.d(TAG, "超过${RECOGNITION_TIMEOUT/1000}秒未检测到语音，停止识别")
isTimeoutPaused = true
playNotification("没有听到您说话，已暂停对话。双击耳机按钮可重新开始。")
- // 停止语音识别并确保资源完全释放
- stopVoiceRecognition()
-
- // 清理识别状态
- isRecognitionActive = false
- hasSpeechDetected = false
+ // 停止语音识别
+ voiceInteractionHandler.stopRecognition()
}
}
}
@@ -340,14 +338,12 @@ class VoiceInteractionService : Service() {
lastKeyEventTime = currentTime
// 停止当前TTS播放
- stopCurrentTTS()
+ voiceInteractionHandler.stopTts()
setPlaybackState(PlaybackStateCompat.STATE_PLAYING)
// 播放提示音
playPrompt("我在!")
- // audioPlayer?.play(R.raw.listening)
-
// 设置为播放状态
setPlaybackState(PlaybackStateCompat.STATE_PAUSED)
@@ -355,221 +351,48 @@ class VoiceInteractionService : Service() {
// 重置超时暂停标志
isTimeoutPaused = false
- // 启动或重置语音识别
- if (!isRecognitionActive) {
+ // 启动语音识别
+ if (!voiceInteractionHandler.isRecognitionActive) {
FileLogger.d(TAG, "语音识别未激活，开始启动")
- // 如果之前是因为超时暂停，重新初始化语音识别组件
- // if (isTimeoutPaused) {
- // Log.d(TAG, "之前因超时暂停，重新初始化Azure服务")
- // restartAsr()
- // }
+ // 通知 Flutter 语音识别已启动
+ notifyRecognitionStarted()
- startVoiceRecognition()
+ // 启动语音识别
+ voiceInteractionHandler.startRecognition()
} else {
FileLogger.d(TAG, "语音识别已激活，更新活动时间")
updateLastActivityTime()
- hasSpeechDetected = false
- }
- }
-
- /**
- * 开始语音识别
- */
- private fun startVoiceRecognition() {
- if (isRecognitionActive) return
-
- // 检查录音权限
- if (!checkRecordAudioPermission()) {
- playNotification("需要录音权限，请在设置中授予权限")
- return
- }
- // 通知 Flutter 语音识别已启动
- notifyRecognitionStarted()
-
- isActive = true
- isRecognitionActive = true
- hasSpeechDetected = false
- updateLastActivityTime()
-
- try {
- azureAsrHelper.startContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
- override fun onRecognizing(recognizing: String, detectedLanguage: String) {
- if (recognizing.isNotEmpty()) {
- hasSpeechDetected = true
- stopCurrentTTS()
- updateLastActivityTime()
- }
- }
-
- override fun onResult(result: String, detectedLanguage: String) {
- if (result.isNotEmpty()) {
- processWithVolcanoAI(result)
- }
-
- // 重置状态，继续识别
- hasSpeechDetected = false
- updateLastActivityTime()
- }
-
- override fun onSessionStarted() {
- updateLastActivityTime()
-
- }
-
- override fun onSessionStopped() {
- isRecognitionActive = false
-
-
- }
-
- override fun onCanceled(reason: String, errorDetails: String) {
- isRecognitionActive = false
-
-
- }
-
- override fun onError(error: String) {
- isRecognitionActive = false
- playNotification("语音识别出错")
-
- }
- })
- } catch (e: Exception) {
- isRecognitionActive = false
- FileLogger.e(TAG, "启动语音识别失败: ${e.message}", e)
- playNotification("启动语音识别失败")
-
-
- }
- }
-
- /**
- * 停止语音识别
- */
- private fun stopVoiceRecognition() {
- if (!isRecognitionActive) return
-
- FileLogger.d(TAG, "停止语音识别")
-
- try {
- azureAsrHelper.stopContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
- override fun onResult(result: String, detectedLanguage: String) {}
- override fun onRecognizing(recognizing: String, detectedLanguage: String) {}
- override fun onSessionStarted() {}
- override fun onSessionStopped() {
- isRecognitionActive = false
- FileLogger.d(TAG, "语音识别会话已停止")
- }
- override fun onCanceled(reason: String, errorDetails: String) {
- isRecognitionActive = false
- FileLogger.d(TAG, "语音识别已取消: $reason")
- }
- override fun onError(error: String) {
- isRecognitionActive = false
- FileLogger.e(TAG, "停止语音识别时出错: $error")
- }
- })
- } catch (e: Exception) {
- FileLogger.e(TAG, "停止语音识别异常: ${e.message}", e)
- // 确保状态一致性
- isRecognitionActive = false
}
-
- setPlaybackState(PlaybackStateCompat.STATE_PAUSED)
- isRecognitionActive = false
- hasSpeechDetected = false
- // 不重置 isTimeoutPaused，保留暂停原因
- }
-
- /**
- * 使用VolcanoAI处理语音识别结果
- */
- private fun processWithVolcanoAI(text: String) {
- // 保存当前用户输入，用于后续同步聊天记录
- currentUserInput = text
-
- Thread {
- try {
- val messages = JSONArray().apply {
- put(volcanoAIService.createUserMessage(text))
- }
-
- val response = volcanoAIService.sendMessage(messages, systemPrompt)
-
- // 播放AI回复
- speakAIResponse(response)
-
- // 同步聊天记录到Flutter端
- notifyChatHistoryUpdated("personal_assistant", text, response)
- } catch (e: Exception) {
- FileLogger.e(TAG, "AI处理出错: ${e.message}")
- playNotification("AI处理出错")
- }
- }.start()
- }
-
- /**
- * 播放AI回复
- */
- private fun speakAIResponse(text: String) {
- isTtsSpeaking = true
- azureTtsHelper.speakText(text, object : AzureTtsHelper.TTSCallback {
- override fun onSuccess(message: String) {
- isTtsSpeaking = false
- updateLastActivityTime()
- }
-
- override fun onError(error: String) {
- isTtsSpeaking = false
- }
- })
}
/**
* 播放提示音
*/
private fun playPrompt(message: String) {
- isTtsSpeaking = true
- azureTtsHelper.speakText(message, object : AzureTtsHelper.TTSCallback {
- override fun onSuccess(message: String) { isTtsSpeaking = false }
- override fun onError(error: String) { isTtsSpeaking = false }
- })
+ voiceInteractionHandler.playTts(message)
}
/**
* 播放通知提示音
*/
private fun playNotification(message: String) {
- isTtsSpeaking = true
// 更新播放状态为播放中，增加接收蓝牙按键事件的几率
setPlaybackState(PlaybackStateCompat.STATE_PLAYING)
// 确保媒体会话处于活跃状态
mediaSession.isActive = true
- azureTtsHelper.speakText(message, object : AzureTtsHelper.TTSCallback {
- override fun onSuccess(message: String) {
- isTtsSpeaking = false
+ voiceInteractionHandler.playTts(message, object : VoiceInteractionHandler.TtsCallback {
+ override fun onComplete() {
setPlaybackState(PlaybackStateCompat.STATE_PAUSED)
}
override fun onError(error: String) {
- isTtsSpeaking = false
setPlaybackState(PlaybackStateCompat.STATE_PAUSED)
}
})
}
- /**
- * 停止当前TTS播放
- */
- private fun stopCurrentTTS() {
- if (isTtsSpeaking) {
- azureTtsHelper.stopSpeaking()
- isTtsSpeaking = false
- }
- }
-
/**
* 更新最后活动时间
*/
@@ -603,15 +426,6 @@ class VoiceInteractionService : Service() {
}
}
- /**
- * 检查录音权限
- */
- private fun checkRecordAudioPermission(): Boolean {
- val permission = android.Manifest.permission.RECORD_AUDIO
- val result = applicationContext.checkCallingOrSelfPermission(permission)
- return result == android.content.pm.PackageManager.PERMISSION_GRANTED
- }
-
/**
* 启动前台服务
*/
@@ -746,38 +560,32 @@ class VoiceInteractionService : Service() {
override fun onBind(intent: Intent?): IBinder? = null
override fun onDestroy() {
- FileLogger.d(TAG, "onDestroy - 语音交互服务正在销毁")
-
- // 释放音频播放器
- audioPlayer?.release()
- audioPlayer = null
+ super.onDestroy()
+ FileLogger.d(TAG, "onDestroy - 语音交互服务即将销毁")
- // 停止监控
+ // 停止服务监控
stopMonitoring()
// 停止语音识别
- if (isRecognitionActive) {
- stopVoiceRecognition()
- }
+ voiceInteractionHandler.stopRecognition()
+
+ // 停止媒体会话
+ mediaSession.release()
+ FileLogger.d(TAG, "媒体会话已释放")
// 停止TTS
- stopCurrentTTS()
+ voiceInteractionHandler.stopTts()
- // 释放 MediaSession
- mediaSession.release()
+ // 释放语音交互处理器资源
+ voiceInteractionHandler.dispose()
- // 释放Azure服务实例
- FileLogger.d(TAG, "释放Azure服务实例")
- azureAsrHelper.dispose()
- azureTtsHelper.dispose()
+ // 关闭音频播放器
+ audioPlayer?.release()
- // 更新服务状态
+ // 设置服务状态
isRunning.set(false)
- // 关闭日志系统
- FileLogger.shutdown()
-
- super.onDestroy()
+ FileLogger.d(TAG, "语音交互服务已销毁")
}
/**
@@ -788,11 +596,11 @@ class VoiceInteractionService : Service() {
FileLogger.d(TAG, "暂停后台语音交互（来自Flutter的请求）")
// 停止当前TTS播放
- stopCurrentTTS()
+ voiceInteractionHandler.stopTts()
// 停止语音识别
- if (isRecognitionActive) {
- stopVoiceRecognition()
+ if (voiceInteractionHandler.isRecognitionActive) {
+ voiceInteractionHandler.stopRecognition()
}
// 设置为暂停状态，但保持服务活跃
@@ -804,10 +612,6 @@ class VoiceInteractionService : Service() {
* 通知 Flutter 端聊天记录已更新
*/
private fun notifyChatHistoryUpdated(agentId: String, userMessage: String, assistantMessage: String) {
- FileLogger.d(TAG, "通知Flutter聊天记录已更新: agentId=$agentId")
- FileLogger.d(TAG, "用户消息: ${userMessage.take(50)}...")
- FileLogger.d(TAG, "助手回复: ${assistantMessage.take(50)}...")
-
// 创建广播 Intent
val intent = Intent(ACTION_CHAT_HISTORY_UPDATED).apply {
putExtra("agentId", agentId)
@@ -818,7 +622,6 @@ class VoiceInteractionService : Service() {
// 发送广播
sendBroadcast(intent)
- FileLogger.d(TAG, "已发送聊天记录广播")
}
/**
@@ -831,6 +634,7 @@ class VoiceInteractionService : Service() {
val intent = Intent(ACTION_RECOGNITION_STARTED).apply {
// 可以添加额外数据
putExtra("timestamp", System.currentTimeMillis())
+ putExtra("packageName", applicationContext.packageName)
}
// 发送广播
@@ -909,5 +713,4 @@ class VoiceInteractionService : Service() {
*/
fun isPlaying() = mediaPlayer?.isPlaying == true
}
-
}
\ No newline at end of file
diff --git a/android/app/src/main/kotlin/com/example/deep_voice/core/utils/FileLogger.kt b/android/app/src/main/kotlin/com/yunqiinnovation/deepsound/core/utils/FileLogger.kt
similarity index 100%
rename from android/app/src/main/kotlin/com/example/deep_voice/core/utils/FileLogger.kt
rename to android/app/src/main/kotlin/com/yunqiinnovation/deepsound/core/utils/FileLogger.kt
diff --git a/android/build.gradle.kts b/android/build.gradle.kts
index 35365df1a..5421f73bd 100644
--- a/android/build.gradle.kts
+++ b/android/build.gradle.kts
@@ -2,6 +2,11 @@ allprojects {
repositories {
google()
mavenCentral()
+ maven { url = uri("https://storage.googleapis.com/download.flutter.io") }
+ // 添加火山引擎Maven仓库
+ maven {
+ url = uri("https://artifact.bytedance.com/repository/Volcengine/")
+ }
}
}
@@ -20,3 +25,14 @@ subprojects {
tasks.register("clean") {
delete(rootProject.layout.buildDirectory)
}
+
+buildscript {
+ repositories {
+ google()
+ mavenCentral()
+ }
+ dependencies {
+ classpath("org.jetbrains.kotlin:kotlin-gradle-plugin:2.1.10")
+ // 其他依赖...
+ }
+}
diff --git a/android/settings.gradle.kts b/android/settings.gradle.kts
index 6296e34ed..e332da5f9 100644
--- a/android/settings.gradle.kts
+++ b/android/settings.gradle.kts
@@ -25,8 +25,15 @@ pluginManagement {
plugins {
id("dev.flutter.flutter-plugin-loader")
id("com.android.application") version "8.7.0" apply false
- id("org.jetbrains.kotlin.android") version "1.8.22" apply false
+ id("org.jetbrains.kotlin.android") version "2.1.10" apply false
}
include(":app")
+include(":azure_speech")
+include(":open_ai_service")
+include(":volcano_speech")
+// 设置azure_speech项目的路径
+project(":azure_speech").projectDir = file("../local_plugins/azure_speech/android")
+project(":open_ai_service").projectDir = file("../local_plugins/open_ai_service/android")
+project(":volcano_speech").projectDir = file("../local_plugins/volcano_speech/android")
diff --git a/azure/LICENSE b/azure/LICENSE
deleted file mode 100644
index d3df29b8a..000000000
--- a/azure/LICENSE
+++ /dev/null
@@ -1,21 +0,0 @@
-MIT License
-
-Copyright (c) 2024 Your Company
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
\ No newline at end of file
diff --git a/azure/README.md b/azure/README.md
deleted file mode 100644
index a19025680..000000000
--- a/azure/README.md
+++ /dev/null
@@ -1,66 +0,0 @@
-# Azure Speech Recognition
-
-A Flutter plugin for Microsoft Azure Speech services, providing both speech recognition (ASR) and text-to-speech (TTS) capabilities.
-
-## Features
-
-- Speech-to-text (Azure Speech Recognition)
-- Text-to-speech (Azure Speech Synthesis)
-- Support for multiple languages
-- Language detection
-- Continuous recognition
-- Streaming synthesis
-
-## Getting Started
-
-### Prerequisites
-
-- Azure Speech service subscription key
-- Azure Speech service region
-
-### Installation
-
-Add this to your package's `pubspec.yaml` file:
-
-```yaml
-dependencies:
- azure_speech_recognition:
- path: ./azure
-```
-
-### Usage
-
-```dart
-import 'package:azure_speech_recognition/azure_speech_recognition.dart';
-
-// Initialize the service
-await AzureSpeechRecognition.initialize(
- subscriptionKey: 'your_subscription_key',
- region: 'your_region',
- supportedLanguages: ['zh-CN', 'en-US'],
-);
-
-// Start continuous recognition
-await AzureSpeechRecognition.startContinuousRecognition();
-
-// Listen for recognition events
-AzureSpeechRecognition.onRecognitionEvent.listen((event) {
- if (event['type'] == 'result') {
- print('Recognized: ${event['text']}');
- print('Detected language: ${event['detectedLanguage']}');
- }
-});
-
-// Stop recognition when done
-await AzureSpeechRecognition.stopContinuousRecognition();
-
-// Speak text
-await AzureSpeechRecognition.speakText('Hello, world!');
-
-// Clean up
-await AzureSpeechRecognition.dispose();
-```
-
-## License
-
-This project is licensed under the MIT License - see the LICENSE file for details.
\ No newline at end of file
diff --git a/azure/ios/Classes/AzureAsrHelper.swift b/azure/ios/Classes/AzureAsrHelper.swift
deleted file mode 100644
index 58b142f60..000000000
--- a/azure/ios/Classes/AzureAsrHelper.swift
+++ /dev/null
@@ -1,757 +0,0 @@
-import Foundation
-import MicrosoftCognitiveServicesSpeech
-import AVFoundation
-import AudioToolbox
-
-/// Azure ASR工具类，负责实现语音识别服务接口
-@available(iOS 13.0, *)
-class AzureAsrHelper: NSObject {
- // MARK: - 属性
-
- /// 事件处理回调
- private var eventHandler: (String, [String: Any]) -> Void
-
- /// 语音配置信息
- private var speechSubscriptionKey: String = ""
- private var serviceRegion: String = ""
-
- /// 语音识别相关
- private var speechConfig: SPXSpeechConfiguration?
- private var recognizer: SPXSpeechRecognizer?
- private var audioConfig: SPXAudioConfiguration?
- private var pushStream: SPXPushAudioInputStream?
-
- /// 音频处理相关
- private var audioProcessor: CustomAudioProcessor?
- private var isProcessingAudio = false
- private var audioProcessingTimer: Timer?
-
- /// 状态标志
- private var isInitialized = false
- private var _isContinuousRecognitionActive = false
-
- /// 当前语言和支持的语言
- private var currentLanguage = "zh-CN"
- private var supportedLanguages: [String] = ["zh-CN", "en-US"]
- private var isAutoDetectLanguage = false
-
- // MARK: - 初始化
-
- init(eventHandler: @escaping (String, [String: Any]) -> Void) {
- self.eventHandler = eventHandler
- super.init()
- }
-
- deinit {
- dispose()
- }
-
- // MARK: - ASR Service 接口实现
-
- /// 初始化语音识别服务
- /// - Parameters:
- /// - speechSubscriptionKey: Azure 语音服务订阅密钥
- /// - serviceRegion: Azure 服务区域 (如 eastasia)
- /// - supportedLanguages: 支持的语言代码数组 (可选)
- /// - Returns: 初始化是否成功
- func initialize(speechSubscriptionKey: String, serviceRegion: String, supportedLanguages: [String]? = nil) -> Bool {
- print("[AzureAsrHelper] 初始化 Azure 语音服务")
-
- // 检查配置是否为空
- if speechSubscriptionKey.isEmpty || serviceRegion.isEmpty {
- print("[AzureAsrHelper] 错误: Azure 配置信息不完整")
- eventHandler("error", ["message": "Azure 配置信息不完整"])
- return false
- }
-
- // 释放之前的资源
- dispose()
-
- // 记录配置信息
- self.speechSubscriptionKey = speechSubscriptionKey
- self.serviceRegion = serviceRegion
-
- // 设置语言
- if let languages = supportedLanguages, !languages.isEmpty {
- self.supportedLanguages = languages
- }
-
- // 根据支持的语言数量决定是否启用自动语言检测
- isAutoDetectLanguage = self.supportedLanguages.count >= 2
-
- // 如果只有一种语言，设置为当前语言
- if !isAutoDetectLanguage && !self.supportedLanguages.isEmpty {
- currentLanguage = self.supportedLanguages[0]
- }
-
- // 创建识别器和设置回调
- if !createRecognizerAndSetupCallbacks() {
- return false
- }
-
- print("[AzureAsrHelper] Azure 语音服务初始化成功")
- isInitialized = true
- return true
- }
-
- /// 创建识别器并设置回调
- private func createRecognizerAndSetupCallbacks() -> Bool {
- // 释放之前的 recognizer
- recognizer = nil
- audioConfig = nil
-
- do {
- // 创建语音配置
- speechConfig = try SPXSpeechConfiguration(subscription: speechSubscriptionKey, region: serviceRegion)
-
- // 设置音频输入参数
- try setupAudioSession()
-
- // 创建自定义推送流，替代默认的麦克风输入
- pushStream = try SPXPushAudioInputStream()
- audioConfig = try SPXAudioConfiguration(streamInput: pushStream!)
-
- // 初始化自定义音频处理器
- audioProcessor = CustomAudioProcessor()
-
- // 设置语言配置
- if isAutoDetectLanguage {
- // 设置自动语言检测
- speechConfig?.setPropertyTo("Continuous", by: SPXPropertyId.speechServiceConnectionLanguageIdMode)
-
- // 创建自动语言检测配置
- let autoDetectSourceLanguageConfig = try SPXAutoDetectSourceLanguageConfiguration(supportedLanguages)
-
- // 创建识别器
- recognizer = try SPXSpeechRecognizer(
- speechConfiguration: speechConfig!,
- autoDetectSourceLanguageConfiguration: autoDetectSourceLanguageConfig,
- audioConfiguration: audioConfig!
- )
- } else {
- // 设置指定的识别语言
- speechConfig?.speechRecognitionLanguage = currentLanguage
-
- // 创建识别器
- recognizer = try SPXSpeechRecognizer(speechConfiguration: speechConfig!, audioConfiguration: audioConfig!)
- }
-
- // 设置所有回调
- setupAllCallbacks()
-
- return true
- } catch {
- print("[AzureAsrHelper] 错误: 创建识别器失败: \(error.localizedDescription)")
- eventHandler("error", ["message": "创建识别器失败: \(error.localizedDescription)"])
- return false
- }
- }
-
- /// 设置音频会话
- private func setupAudioSession() throws {
- let audioSession = AVAudioSession.sharedInstance()
-
- // 使用playAndRecord类别允许同时录音和播放
- try audioSession.setCategory(.playAndRecord,
- mode: .voiceChat, // 使用voiceChat模式能够更好地支持回音消除
- options: [.allowBluetooth, .defaultToSpeaker, .allowAirPlay, .mixWithOthers])
-
- // 设置首选的输入和输出
- let currentRoute = audioSession.currentRoute
-
- // 获取当前是否连接了耳机或外部麦克风
- let hasHeadphones = currentRoute.outputs.contains {
- $0.portType == .headphones || $0.portType == .bluetoothA2DP || $0.portType == .bluetoothHFP
- }
-
- // 如果没有耳机，明确启用内置麦克风和扬声器的回音消除
- if !hasHeadphones {
- try audioSession.setMode(.voiceChat) // 语音聊天模式有更强的回音消除
-
- // 启用回音消除和噪声抑制
- try audioSession.setInputGain(0.8) // 适当降低输入增益以减少扬声器音频被麦克风捕获的可能性
- } else {
- // 耳机模式，可以使用不同的设置
- try audioSession.setMode(.voiceChat)
- try audioSession.setInputGain(1.0)
- }
-
- // 设置合适的采样率
- try audioSession.setPreferredSampleRate(16000.0) // Azure语音识别推荐的采样率
- try audioSession.setPreferredIOBufferDuration(0.01) // 较小的缓冲区大小以减少延迟
-
- // 激活音频会话
- try audioSession.setActive(true, options: .notifyOthersOnDeactivation)
-
- print("[AzureAsrHelper] 音频会话配置成功，已启用回音消除")
- }
-
- /// 设置所有回调
- private func setupAllCallbacks() {
- guard let recognizer = recognizer else { return }
-
- // 最终识别结果
- recognizer.addRecognizedEventHandler { [weak self] _, event in
- guard let self = self else { return }
-
- if event.result.reason == SPXResultReason.recognizedSpeech {
- let detectedLanguage = self.getDetectedLanguage(from: event.result)
- print("[AzureAsrHelper] 识别结果: \(event.result.text ?? ""), 语言: \(detectedLanguage)")
- self.eventHandler("result", [
- "text": event.result.text ?? "",
- "detectedLanguage": detectedLanguage
- ])
- }
- }
-
- // 识别中事件
- recognizer.addRecognizingEventHandler { [weak self] _, event in
- guard let self = self else { return }
-
- if event.result.reason == SPXResultReason.recognizingSpeech {
- let detectedLanguage = self.getDetectedLanguage(from: event.result)
- // print("[AzureAsrHelper] 识别中: \(event.result.text ?? ""), 语言: \(detectedLanguage)")
- self.eventHandler("recognizing", [
- "text": event.result.text ?? "",
- "detectedLanguage": detectedLanguage
- ])
- }
- }
-
- // 会话事件
- recognizer.addSessionStartedEventHandler { [weak self] _, _ in
- guard let self = self else { return }
-
- print("[AzureAsrHelper] 识别会话已开始")
- self._isContinuousRecognitionActive = true
- self.eventHandler("sessionStarted", [:])
- }
-
- recognizer.addSessionStoppedEventHandler { [weak self] _, _ in
- guard let self = self else { return }
-
- print("[AzureAsrHelper] 识别会话已结束")
- self._isContinuousRecognitionActive = false
- self.eventHandler("sessionStopped", [:])
- }
-
- // 取消事件
- recognizer.addCanceledEventHandler { [weak self] _, event in
- guard let self = self else { return }
-
- let reason = event.reason.rawValue
- let errorDetails = event.errorDetails ?? "未知错误"
-
- print("[AzureAsrHelper] 识别取消: \(errorDetails)")
-
- self.eventHandler("canceled", [
- "reason": reason,
- "errorDetails": errorDetails
- ])
-
- self._isContinuousRecognitionActive = false
- }
- }
-
- /// 执行一次性语音识别
- /// - Returns: 是否成功启动识别
- func recognizeOnce() -> Bool {
- if !isInitialized {
- print("[AzureAsrHelper] 错误: 语音服务未初始化")
- eventHandler("error", ["message": "语音服务未初始化"])
- return false
- }
-
- // 如果正在连续识别，先停止
- if _isContinuousRecognitionActive {
- stopContinuousRecognition()
- }
-
- // 确保识别器已创建
- if recognizer == nil && !createRecognizerAndSetupCallbacks() {
- return false
- }
-
- do {
- // 启动音频处理
- startAudioProcessing()
-
- // 通知会话开始
- eventHandler("sessionStarted", [:])
-
- // 执行识别
- try recognizer?.recognizeOnceAsync { [weak self] result in
- guard let self = self else { return }
-
- // 停止音频处理
- self.stopAudioProcessing()
-
- if result.reason == SPXResultReason.recognizedSpeech {
- let detectedLanguage = self.getDetectedLanguage(from: result)
- self.eventHandler("result", [
- "text": result.text ?? "",
- "detectedLanguage": detectedLanguage
- ])
- } else if result.reason == SPXResultReason.noMatch {
- print("[AzureAsrHelper] 无匹配结果")
- self.eventHandler("noMatch", [:])
- } else if result.reason == SPXResultReason.canceled {
- do {
- let details = try SPXCancellationDetails(fromCanceledRecognitionResult: result)
- let errorDetails = details.errorDetails ?? "未知错误"
- self.eventHandler("error", ["message": "识别取消: \(errorDetails)"])
- } catch {
- print("[AzureAsrHelper] 错误: 获取取消详情失败: \(error.localizedDescription)")
- self.eventHandler("error", ["message": "识别取消，无法获取详细原因"])
- }
- }
- }
-
- return true
- } catch {
- print("[AzureAsrHelper] 错误: 识别异常: \(error.localizedDescription)")
- eventHandler("error", ["message": "识别异常: \(error.localizedDescription)"])
- stopAudioProcessing()
- return false
- }
- }
-
- /// 开始连续语音识别
- /// - Returns: 是否成功启动识别
- func startContinuousRecognition() -> Bool {
- if !isInitialized {
- print("[AzureAsrHelper] 错误: 语音服务未初始化")
- eventHandler("error", ["message": "语音服务未初始化"])
- return false
- }
-
- // 如果已经在进行连续识别，先停止
- if _isContinuousRecognitionActive {
- stopContinuousRecognition()
- }
-
- // 确保识别器已创建
- if recognizer == nil && !createRecognizerAndSetupCallbacks() {
- return false
- }
-
- // 重新确保音频设置正确
- do {
- try setupAudioSession()
- } catch {
- print("[AzureAsrHelper] 警告: 设置音频会话失败: \(error.localizedDescription)")
- }
-
- do {
- // 启动音频处理
- startAudioProcessing()
-
- // 启动连续识别
- try recognizer?.startContinuousRecognition()
- _isContinuousRecognitionActive = true
-
- print("[AzureAsrHelper] 连续识别开始")
- return true
- } catch {
- print("[AzureAsrHelper] 错误: 开始连续识别失败: \(error.localizedDescription)")
- eventHandler("error", ["message": "开始连续识别失败: \(error.localizedDescription)"])
- _isContinuousRecognitionActive = false
- stopAudioProcessing()
- return false
- }
- }
-
- /// 停止连续语音识别
- /// - Returns: 是否成功停止识别
- func stopContinuousRecognition() -> Bool {
- // 停止音频处理
- stopAudioProcessing()
-
- if !_isContinuousRecognitionActive || recognizer == nil {
- return true
- }
-
- do {
- try recognizer?.stopContinuousRecognition()
- _isContinuousRecognitionActive = false
- print("[AzureAsrHelper] 连续识别已停止")
- return true
- } catch {
- print("[AzureAsrHelper] 错误: 停止连续识别失败: \(error.localizedDescription)")
- eventHandler("error", ["message": "停止连续识别失败: \(error.localizedDescription)"])
- _isContinuousRecognitionActive = false
- return false
- }
- }
-
- /// 检查连续识别是否活跃
- /// - Returns: 连续识别是否处于活跃状态
- func isContinuousRecognitionActive() -> Bool {
- return _isContinuousRecognitionActive
- }
-
- /// 释放资源
- func dispose() {
- print("[AzureAsrHelper] 释放资源")
-
- // 停止音频处理
- stopAudioProcessing()
-
- // 停止连续识别
- if _isContinuousRecognitionActive {
- stopContinuousRecognition()
- }
-
- // 释放音频会话
- do {
- try AVAudioSession.sharedInstance().setActive(false, options: .notifyOthersOnDeactivation)
- } catch {
- print("[AzureAsrHelper] 警告: 释放音频会话失败: \(error.localizedDescription)")
- }
-
- // 释放资源
- recognizer = nil
- speechConfig = nil
- audioConfig = nil
- pushStream = nil
- audioProcessor = nil
-
- // 重置状态
- _isContinuousRecognitionActive = false
- isInitialized = false
- }
-
- /// 从结果中获取检测到的语言
- private func getDetectedLanguage(from result: SPXSpeechRecognitionResult) -> String {
- if isAutoDetectLanguage {
- do {
- let langResult = try SPXAutoDetectSourceLanguageResult(result)
- return langResult.language ?? currentLanguage
- } catch {
- print("[AzureAsrHelper] 错误: 获取检测到的语言失败: \(error.localizedDescription)")
- return currentLanguage
- }
- } else {
- return currentLanguage
- }
- }
-
- // MARK: - 音频处理
-
- /// 开始音频处理
- private func startAudioProcessing() {
- guard !isProcessingAudio, let audioProcessor = audioProcessor else { return }
-
- isProcessingAudio = true
-
- // 启动音频处理器
- if !audioProcessor.startRecord() {
- print("[AzureAsrHelper] 错误: 启动音频处理器失败")
- eventHandler("error", ["message": "启动音频处理器失败"])
- return
- }
-
- // 启动音频处理定时器
- audioProcessingTimer = Timer.scheduledTimer(withTimeInterval: 0.08, repeats: true) { [weak self] _ in
- guard let self = self, self.isProcessingAudio, let processor = self.audioProcessor, let stream = self.pushStream else {
- return
- }
-
- // 读取处理后的音频数据
- var bytes = [UInt8](repeating: 0, count: 2560)
- let bytesRead = processor.read(bytes: &bytes)
-
- if bytesRead > 0 {
- // 推送数据到Azure语音服务
- let data = Data(bytes: bytes, count: bytesRead)
- stream.write(data)
-
- // 通知音频数据可用
- self.eventHandler("audioData", ["data": bytes])
- }
- }
-
- print("[AzureAsrHelper] 音频处理已启动")
- }
-
- /// 停止音频处理
- private func stopAudioProcessing() {
- // 停止定时器
- audioProcessingTimer?.invalidate()
- audioProcessingTimer = nil
-
- // 停止音频处理器
- audioProcessor?.stopRecord()
-
- isProcessingAudio = false
- print("[AzureAsrHelper] 音频处理已停止")
- }
-}
-
-// MARK: - 自定义音频处理器
-
-@available(iOS 13.0, *)
-class CustomAudioProcessor: NSObject {
- // 音频单元
- private var ioUnit: AudioUnit?
-
- // 音频格式
- private var audioFormat: AudioStreamBasicDescription
-
- // 音频缓冲
- private var audioBufferList: AudioBufferList
- private var audioList: [Float] = []
- private let audioListQueue = DispatchQueue(label: "audioListQueue")
-
- // 回音消除状态
- private var isEchoCancellationEnabled = true
-
- override init() {
- // 设置音频格式 - 16kHz, 16位, 单声道
- audioFormat = AudioStreamBasicDescription(
- mSampleRate: 16000.0,
- mFormatID: kAudioFormatLinearPCM,
- mFormatFlags: kAudioFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked,
- mBytesPerPacket: 2,
- mFramesPerPacket: 1,
- mBytesPerFrame: 2,
- mChannelsPerFrame: 1,
- mBitsPerChannel: 16,
- mReserved: 0
- )
-
- // 初始化音频缓冲
- audioBufferList = AudioBufferList(
- mNumberBuffers: 1,
- mBuffers: AudioBuffer(
- mNumberChannels: 1,
- mDataByteSize: 4096,
- mData: malloc(4096)
- )
- )
-
- super.init()
- }
-
- deinit {
- stopRecord()
- free(audioBufferList.mBuffers.mData)
- }
-
- /// 启动音频处理
- /// - Returns: 是否成功启动
- func startRecord() -> Bool {
- print("[CustomAudioProcessor] 配置音频单元")
-
- // 创建音频组件描述 - 使用VoiceProcessingIO类型获取回音消除
- var ioUnitDescription = AudioComponentDescription(
- componentType: kAudioUnitType_Output,
- componentSubType: kAudioUnitSubType_VoiceProcessingIO,
- componentManufacturer: kAudioUnitManufacturer_Apple,
- componentFlags: 0,
- componentFlagsMask: 0
- )
-
- // 查找音频组件
- guard let ioUnitRef = AudioComponentFindNext(nil, &ioUnitDescription) else {
- print("[CustomAudioProcessor] 错误: 未找到音频组件")
- return false
- }
-
- // 创建音频单元实例
- if checkError(AudioComponentInstanceNew(ioUnitRef, &ioUnit), "创建音频单元") {
- ioUnit = nil
- return false
- }
-
- // 启用输入端口
- var enableInput: UInt32 = 1
- let kInputBus: AudioUnitElement = 1
- let kOutputBus: AudioUnitElement = 0
- if checkError(AudioUnitSetProperty(ioUnit!, kAudioOutputUnitProperty_EnableIO,
- kAudioUnitScope_Input, kInputBus, &enableInput,
- UInt32(MemoryLayout.size)), "启用输入端口") {
- return false
- }
-
- // 禁用输出端口 (我们只需要输入)
- var enableOutput: UInt32 = 0
- if checkError(AudioUnitSetProperty(ioUnit!, kAudioOutputUnitProperty_EnableIO,
- kAudioUnitScope_Output, kOutputBus,
- &enableOutput, UInt32(MemoryLayout.size)), "禁用输出端口") {
- return false
- }
-
- // 设置缓冲区分配标志
- var flag: UInt32 = 0
- if checkError(AudioUnitSetProperty(ioUnit!, kAudioUnitProperty_ShouldAllocateBuffer,
- kAudioUnitScope_Output, kInputBus, &flag, UInt32(MemoryLayout.size)), "设置缓冲区分配标志") {
- return false
- }
-
- // 设置音频格式
- let size = UInt32(MemoryLayout.size)
- if checkError(AudioUnitSetProperty(ioUnit!, kAudioUnitProperty_StreamFormat,
- kAudioUnitScope_Output, kInputBus, &audioFormat, size), "设置输入总线输出范围的流格式") {
- return false
- }
-
- if checkError(AudioUnitSetProperty(ioUnit!, kAudioUnitProperty_StreamFormat,
- kAudioUnitScope_Input, kOutputBus, &audioFormat, size), "设置输出总线输入范围的流格式") {
- return false
- }
-
- // 启用回音消除
- if isEchoCancellationEnabled {
- var echoCancellation: UInt32 = 1
- AudioUnitSetProperty(ioUnit!, kAUVoiceIOProperty_BypassVoiceProcessing,
- kAudioUnitScope_Global, 0, &echoCancellation, UInt32(MemoryLayout.size))
- }
-
- // 设置输入回调 - 当有新音频数据时调用
- var inputCallback = AURenderCallbackStruct(
- inputProc: CustomAudioProcessor.onAudioDataAvailable,
- inputProcRefCon: UnsafeMutableRawPointer(Unmanaged.passUnretained(self).toOpaque())
- )
-
- if checkError(AudioUnitSetProperty(ioUnit!,
- kAudioOutputUnitProperty_SetInputCallback,
- kAudioUnitScope_Global, kInputBus,
- &inputCallback, UInt32(MemoryLayout.size)), "设置输入回调") {
- return false
- }
-
- // 初始化音频单元
- var hasError = checkError(AudioUnitInitialize(ioUnit!), "初始化音频单元")
- while hasError {
- Thread.sleep(forTimeInterval: 0.1)
- hasError = checkError(AudioUnitInitialize(ioUnit!), "初始化音频单元")
- }
-
- // 启动音频单元
- hasError = checkError(AudioOutputUnitStart(ioUnit!), "启动音频单元")
-
- print("[CustomAudioProcessor] 音频处理器已启动，回音消除\(isEchoCancellationEnabled ? "已启用" : "已禁用")")
- return !hasError
- }
-
- /// 停止音频处理
- func stopRecord() {
- print("[CustomAudioProcessor] 停止音频处理器")
-
- if let ioUnit = ioUnit {
- // 停止音频单元
- _ = checkError(AudioOutputUnitStop(ioUnit), "停止音频单元")
-
- // 关闭音频单元
- _ = checkError(AudioUnitUninitialize(ioUnit), "反初始化音频单元")
- _ = checkError(AudioComponentInstanceDispose(ioUnit), "释放音频单元")
-
- self.ioUnit = nil
- }
-
- // 清空音频数据缓冲
- audioListQueue.sync {
- audioList.removeAll()
- }
- }
-
- /// 音频数据回调 - 当有新的音频数据可用时调用
- private static let onAudioDataAvailable: AURenderCallback = { inRefCon, ioActionFlags, inTimeStamp, inBusNumber, inNumberFrames, ioData in
- // 获取实例
- let processor = Unmanaged.fromOpaque(inRefCon).takeUnretainedValue()
-
- // 计算预期数据大小
- let expectedDataByteSize = inNumberFrames * processor.audioFormat.mBytesPerFrame
-
- // 确保缓冲区足够大
- if processor.audioBufferList.mBuffers.mDataByteSize < expectedDataByteSize {
- processor.audioBufferList.mBuffers.mData = realloc(processor.audioBufferList.mBuffers.mData, Int(expectedDataByteSize))
- processor.audioBufferList.mBuffers.mDataByteSize = expectedDataByteSize
- }
-
- // 渲染音频数据
- let status = processor.checkOSStatus(AudioUnitRender(processor.ioUnit!, ioActionFlags, inTimeStamp,
- inBusNumber, inNumberFrames, &processor.audioBufferList),
- "渲染音频数据")
-
- // 将Int16数据转换为浮点数据进行处理
- var audioDataFloat = [Float](repeating: 0.0, count: Int(inNumberFrames))
- let buffer = processor.audioBufferList.mBuffers
- let bufferData = buffer.mData!.assumingMemoryBound(to: Int16.self)
-
- for j in 0...size)) {
- // 归一化到[-1.0, 1.0]范围
- audioDataFloat[j] = Float(bufferData[j]) / 32768.0
- }
-
- // 应用附加处理 (如有需要)
- // processor.applyAdditionalProcessing(&audioDataFloat)
-
- // 保存处理后的数据
- if status == noErr {
- processor.audioListQueue.async {
- processor.audioList.append(contentsOf: audioDataFloat)
- }
- }
-
- return status
- }
-
- /// 读取处理后的音频数据
- /// - Parameter bytes: 输出字节数组
- /// - Returns: 读取的字节数
- func read(bytes: inout [UInt8]) -> Int {
- return audioListQueue.sync {
- // 如果没有数据，返回0
- if audioList.isEmpty {
- return 0
- }
-
- // 确保有足够的数据 (至少1280个样本)
- if audioList.count < 1280 {
- return 0
- }
-
- // 读取一帧数据 (1280个样本)
- let frameLength = 1280
- let buffer = Array(audioList.prefix(frameLength))
- audioList.removeFirst(frameLength)
-
- // 将浮点数据转回Int16格式
- var int16Data = buffer.map { Int16($0 * 32767) }
-
- // 转换为字节数组
- let data = Data(buffer: UnsafeBufferPointer(start: &int16Data, count: int16Data.count))
- bytes = [UInt8](data)
-
- // 每个样本2字节 (16位PCM)
- return frameLength * 2
- }
- }
-
- /// 检查错误并打印日志
- /// - Parameters:
- /// - status: 操作状态
- /// - operation: 操作描述
- /// - Returns: 是否发生错误
- private func checkError(_ status: OSStatus, _ operation: String) -> Bool {
- if status != noErr {
- print("[CustomAudioProcessor] 错误: \(operation)失败: \(status)")
- return true
- }
- return false
- }
-
- /// 检查OSStatus并返回状态
- /// - Parameters:
- /// - status: 操作状态
- /// - operation: 操作描述
- /// - Returns: 原始状态
- private func checkOSStatus(_ status: OSStatus, _ operation: String) -> OSStatus {
- if status != noErr {
- print("[CustomAudioProcessor] 错误: \(operation)失败: \(status)")
- }
- return status
- }
-}
\ No newline at end of file
diff --git a/azure/ios/Classes/AzureSpeechRecognitionPlugin.swift b/azure/ios/Classes/AzureSpeechRecognitionPlugin.swift
deleted file mode 100644
index 141b77ef8..000000000
--- a/azure/ios/Classes/AzureSpeechRecognitionPlugin.swift
+++ /dev/null
@@ -1,18 +0,0 @@
-import Flutter
-import UIKit
-
-public class AzureSpeechRecognitionPlugin: NSObject, FlutterPlugin {
- public static func register(with registrar: FlutterPluginRegistrar) {
- if #available(iOS 13.0, *) {
- SwiftAzureSpeechRecognitionPlugin.register(with: registrar)
- } else {
- // 如果低于iOS 13.0，返回不支持的错误
- let channel = FlutterMethodChannel(name: "com.deep_voice.azure_asr", binaryMessenger: registrar.messenger())
- channel.setMethodCallHandler { (call, result) in
- result(FlutterError(code: "UNSUPPORTED",
- message: "需要iOS 13.0及以上系统",
- details: nil))
- }
- }
- }
-}
\ No newline at end of file
diff --git a/azure/ios/Classes/AzureTtsHelper.swift b/azure/ios/Classes/AzureTtsHelper.swift
deleted file mode 100644
index 589df617c..000000000
--- a/azure/ios/Classes/AzureTtsHelper.swift
+++ /dev/null
@@ -1,427 +0,0 @@
-import Foundation
-import MicrosoftCognitiveServicesSpeech
-import AVFoundation
-
-/// Azure TTS工具类，负责实现TTS服务接口
-@available(iOS 13.0, *)
-class AzureTtsHelper: NSObject {
- // MARK: - 属性
-
- /// 事件处理回调
- private var eventHandler: (String, [String: Any]) -> Void
-
- /// 语音配置信息
- private var speechSubscriptionKey: String = ""
- private var serviceRegion: String = ""
-
- /// 语音合成配置
- private var speechConfig: SPXSpeechConfiguration?
-
- /// 语音合成器
- private var synthesizer: SPXSpeechSynthesizer?
-
- /// 是否初始化成功
- private var isInitialized = false
-
- /// 当前是否正在播放
- private var _isSpeaking = false
-
- /// 音频会话配置
- private var isAudioSessionConfigured = false
-
- // MARK: - 语音设置
-
- /// 当前语音
- private var currentVoice = "zh-CN-XiaoxiaoNeural"
-
- /// 支持的语音映射
- private var voiceMap: [String: String] = [
- "zh-CN": "zh-CN-XiaoxiaoNeural",
- "en-US": "en-US-JennyNeural",
- "ja-JP": "ja-JP-NanamiNeural",
- "ko-KR": "ko-KR-SunHiNeural",
- "zh-TW": "zh-TW-HsiaoChenNeural",
- "zh-HK": "zh-HK-HiuMaanNeural"
- ]
-
- /// 当前语音合成参数
- private var currentSpeechRate = "0%"
- private var currentPitch = "0%"
- private var currentVolume = "100%"
-
- // MARK: - 初始化
-
- init(eventHandler: @escaping (String, [String: Any]) -> Void) {
- self.eventHandler = eventHandler
- super.init()
- }
-
- deinit {
- dispose()
- }
-
- // MARK: - TTS 接口实现
-
- /// 初始化语音合成服务
- /// - Parameters:
- /// - speechSubscriptionKey: Azure 语音服务订阅密钥
- /// - serviceRegion: Azure 服务区域 (如 eastasia)
- /// - language: 语言代码 (默认 zh-CN)
- /// - Returns: 初始化是否成功
- func initialize(speechSubscriptionKey: String, serviceRegion: String, language: String = "zh-CN") -> Bool {
- print("[AzureTtsHelper] 初始化语音合成服务")
-
- // 检查配置是否为空
- if speechSubscriptionKey.isEmpty || serviceRegion.isEmpty {
- print("[AzureTtsHelper] 错误: Azure 配置信息不完整")
- eventHandler("error", ["error": "Azure 配置信息不完整"])
- return false
- }
-
- // 释放之前的资源
- dispose()
-
- // 记录配置信息
- self.speechSubscriptionKey = speechSubscriptionKey
- self.serviceRegion = serviceRegion
-
- // 配置音频会话
- if !configureAudioSession() {
- print("[AzureTtsHelper] 警告: 音频会话配置失败，将尝试继续初始化")
- }
-
- do {
- // 创建语音配置
- speechConfig = try SPXSpeechConfiguration(subscription: speechSubscriptionKey, region: serviceRegion)
-
- // 设置默认语音
- let defaultVoice = getDefaultVoiceForLanguage(language)
- currentVoice = defaultVoice
- speechConfig?.speechSynthesisVoiceName = defaultVoice
-
- // 创建语音合成器
- synthesizer = try SPXSpeechSynthesizer(speechConfig!)
-
- // 设置事件处理器
- setupSynthesizerEvents()
-
- isInitialized = true
- print("[AzureTtsHelper] TTS 引擎初始化成功")
-
- return true
- } catch {
- print("[AzureTtsHelper] 错误: 初始化语音合成服务失败: \(error.localizedDescription)")
- eventHandler("error", ["error": "初始化语音合成服务失败: \(error.localizedDescription)"])
- return false
- }
- }
-
- /// 配置音频会话
- private func configureAudioSession() -> Bool {
- let audioSession = AVAudioSession.sharedInstance()
- do {
- // 使用playback类别，但支持混合和空中播放
- try audioSession.setCategory(.playback,
- mode: .spokenAudio,
- options: [.mixWithOthers, .allowAirPlay, .duckOthers])
-
- // 根据设备类型选择最佳配置
- let currentRoute = audioSession.currentRoute
- let hasHeadphones = currentRoute.outputs.contains {
- $0.portType == .headphones || $0.portType == .bluetoothA2DP || $0.portType == .bluetoothHFP
- }
-
- // 优化音频路由
- if hasHeadphones {
- // 耳机模式，使用默认设置
- try audioSession.setPreferredIOBufferDuration(0.005) // 较小的缓冲区大小以减少延迟
- } else {
- // 扬声器模式
- try audioSession.setPreferredIOBufferDuration(0.005)
- }
-
- // 避免完全激活音频会话，因为ASR可能已经激活
- // 这里使用setActive(false)是为了不与ASR冲突
- if !audioSession.isOtherAudioPlaying {
- try audioSession.setActive(true, options: .notifyOthersOnDeactivation)
- }
-
- isAudioSessionConfigured = true
- print("[AzureTtsHelper] 音频会话配置成功")
- return true
- } catch {
- print("[AzureTtsHelper] 警告: 音频会话配置失败: \(error.localizedDescription)")
- isAudioSessionConfigured = false
- return false
- }
- }
-
- /// 设置语音
- /// - Parameter voiceName: 语音名称 (如 "zh-CN-XiaoxiaoNeural")
- /// - Returns: 设置是否成功
- func setVoice(voiceName: String) -> Bool {
- if !isInitialized {
- print("[AzureTtsHelper] 错误: TTS 引擎尚未初始化")
- eventHandler("error", ["error": "TTS 引擎尚未初始化"])
- return false
- }
-
- if voiceName.isEmpty {
- print("[AzureTtsHelper] 错误: 声音名称为空")
- eventHandler("error", ["error": "声音名称不能为空"])
- return false
- }
-
- if voiceName == currentVoice {
- print("[AzureTtsHelper] 已设置语音: \(voiceName)")
- return true
- }
-
- print("[AzureTtsHelper] 设置声音: \(voiceName)")
- currentVoice = voiceName
-
- // 更新语音配置
- if let speechConfig = speechConfig {
- speechConfig.speechSynthesisVoiceName = voiceName
- return true
- }
-
- return false
- }
-
- /// 设置语音合成参数
- /// - Parameters:
- /// - rate: 语速，范围 -100 到 100，默认为 0
- /// - pitch: 音调，范围 -100 到 100，默认为 0
- /// - volume: 音量，范围 0 到 100，默认为 100
- /// - Returns: 是否设置成功
- func setSpeechParams(rate: Int = 0, pitch: Int = 0, volume: Int = 100) -> Bool {
- if !isInitialized {
- print("[AzureTtsHelper] 错误: TTS 引擎尚未初始化")
- eventHandler("error", ["error": "TTS 引擎尚未初始化"])
- return false
- }
-
- // 转换参数格式
- currentSpeechRate = formatRateParam(rate)
- currentPitch = formatPitchParam(pitch)
- currentVolume = formatVolumeParam(volume)
-
- print("[AzureTtsHelper] 已设置语音参数: 语速=\(currentSpeechRate), 音调=\(currentPitch), 音量=\(currentVolume)")
- return true
- }
-
- /// 合成文本为语音并播放
- /// - Parameter text: 要合成的文本
- /// - Returns: 操作是否成功启动
- func speakText(text: String) -> Bool {
- if !isInitialized {
- print("[AzureTtsHelper] 错误: TTS 引擎尚未初始化")
- eventHandler("error", ["error": "TTS 引擎尚未初始化"])
- return false
- }
-
- if text.isEmpty {
- print("[AzureTtsHelper] 警告: 要播放的文本为空")
- return true
- }
-
- // 确保音频会话已配置
- if !isAudioSessionConfigured {
- _ = configureAudioSession()
- }
-
- print("[AzureTtsHelper] 开始语音合成: \(text.prefix(50))...")
-
- // 生成SSML
- let ssml = generateSsml(text: text)
-
- // 直接进行SSML合成
- return speakSsmlInternal(text: ssml)
- }
-
- /// 内部SSML合成和播放
- private func speakSsmlInternal(text: String) -> Bool {
- guard let synthesizer = synthesizer else {
- print("[AzureTtsHelper] 错误: 合成器未初始化")
- eventHandler("error", ["error": "合成器未初始化"])
- return false
- }
-
- _isSpeaking = true
- eventHandler("started", [:])
-
- Task {
- do {
- // 使用异步方法进行合成并直接播放
- _ = try await synthesizer.startSpeakingSsml(text)
-
- } catch {
- print("[AzureTtsHelper] 错误: 语音合成失败: \(error.localizedDescription)")
- DispatchQueue.main.async {
- self._isSpeaking = false
- self.eventHandler("error", ["error": "语音合成失败: \(error.localizedDescription)"])
- }
- }
- }
-
- return true
- }
-
- /// 停止当前语音合成
- /// - Returns: 操作是否成功
- func stopSpeaking() -> Bool {
- if !isInitialized || !_isSpeaking {
- return true
- }
-
- // 停止合成
- do {
- try synthesizer?.stopSpeaking()
- _isSpeaking = false
- eventHandler("canceled", [:])
- print("[AzureTtsHelper] 已停止语音合成")
- return true
- } catch {
- print("[AzureTtsHelper] 错误: 停止语音合成失败: \(error.localizedDescription)")
- eventHandler("error", ["error": "停止语音合成失败: \(error.localizedDescription)"])
- return false
- }
- }
-
- /// 检查是否正在播放
- /// - Returns: 当前是否正在播放语音
- func isSpeaking() -> Bool {
- return _isSpeaking
- }
-
- /// 释放资源
- func dispose() {
- try? stopSpeaking()
-
- // 释放合成器和配置
- synthesizer = nil
- speechConfig = nil
-
- isInitialized = false
- _isSpeaking = false
- isAudioSessionConfigured = false
- print("[AzureTtsHelper] TTS 引擎已释放")
- }
-
- // MARK: - 私有辅助方法
-
- /// 设置合成器事件处理
- private func setupSynthesizerEvents() {
- guard let synthesizer = synthesizer else { return }
-
- // 添加书签到达事件处理
- synthesizer.addBookmarkReachedEventHandler { _, e in
- print("[AzureTtsHelper] 书签事件: 音频偏移: \((e.audioOffset + 5000) / 10000)ms, 文本: \"\(e.text)\"")
- }
-
- // 合成完成事件
- synthesizer.addSynthesisCompletedEventHandler { [weak self] _, e in
- guard let self = self else { return }
- print("[AzureTtsHelper] 语音合成完成: 音频持续时间: \(e.result.audioDuration)")
- DispatchQueue.main.async {
- self._isSpeaking = false
- self.eventHandler("completed", [:])
- }
- }
-
- // 合成取消事件
- synthesizer.addSynthesisCanceledEventHandler { [weak self] _, e in
- guard let self = self else { return }
-
- let result = e.result
- do {
- let cancellationDetails = try SPXSpeechSynthesisCancellationDetails(fromCanceledSynthesisResult: result)
- print("[AzureTtsHelper] 语音合成取消: 原因: \(cancellationDetails.reason)")
-
- if cancellationDetails.reason == SPXCancellationReason.error {
- print("[AzureTtsHelper] 错误代码: \(cancellationDetails.errorCode)")
- print("[AzureTtsHelper] 错误详情: \(cancellationDetails.errorDetails ?? "未知")")
- }
-
- DispatchQueue.main.async {
- self._isSpeaking = false
- self.eventHandler("error", ["error": "语音合成取消: \(cancellationDetails.errorDetails ?? "未知错误")"])
- }
- } catch {
- print("[AzureTtsHelper] 获取取消详情时出错: \(error)")
-
- DispatchQueue.main.async {
- self._isSpeaking = false
- self.eventHandler("error", ["error": "语音合成被取消"])
- }
- }
- }
-
- // 合成开始事件
- synthesizer.addSynthesisStartedEventHandler { _, _ in
- // print("[AzureTtsHelper] 语音合成开始")
- }
-
- // 合成中事件
- synthesizer.addSynthesizingEventHandler { _, _ in
- // print("[AzureTtsHelper] 语音合成中")
- }
- }
-
- /// 生成 SSML 文本
- private func generateSsml(text: String) -> String {
- return """
-
-
-
- \(text)
-
-
-
- """
- }
-
- /// 格式化语速参数
- private func formatRateParam(_ rate: Int) -> String {
- let clampedRate = rate.clamp(min: -100, max: 100)
- if clampedRate == 0 {
- return "0%"
- } else if clampedRate < 0 {
- return "\(Int(Double(clampedRate) * 0.9))%"
- } else {
- return "+\(clampedRate)%"
- }
- }
-
- /// 格式化音调参数
- private func formatPitchParam(_ pitch: Int) -> String {
- let clampedPitch = pitch.clamp(min: -100, max: 100)
- if clampedPitch == 0 {
- return "0%"
- } else {
- return "\(Int(Double(clampedPitch) * 0.5))%"
- }
- }
-
- /// 格式化音量参数
- private func formatVolumeParam(_ volume: Int) -> String {
- let clampedVolume = volume.clamp(min: 0, max: 100)
- return "\(clampedVolume)%"
- }
-
- /// 获取指定语言的默认语音
- private func getDefaultVoiceForLanguage(_ language: String) -> String {
- return voiceMap[language] ?? "zh-CN-XiaoxiaoNeural"
- }
-}
-
-// MARK: - 扩展
-
-extension Int {
- func clamp(min: Int, max: Int) -> Int {
- if self < min { return min }
- if self > max { return max }
- return self
- }
-}
\ No newline at end of file
diff --git a/azure/ios/Classes/SwiftAzureSpeechRecognitionPlugin.swift b/azure/ios/Classes/SwiftAzureSpeechRecognitionPlugin.swift
deleted file mode 100644
index ea393e215..000000000
--- a/azure/ios/Classes/SwiftAzureSpeechRecognitionPlugin.swift
+++ /dev/null
@@ -1,259 +0,0 @@
-import Flutter
-import UIKit
-import MicrosoftCognitiveServicesSpeech
-import AVFoundation
-
-@available(iOS 13.0, *)
-public class SwiftAzureSpeechRecognitionPlugin: NSObject, FlutterPlugin {
- private var azureChannel: FlutterMethodChannel
- private var ttsChannel: FlutterMethodChannel
- private var asrHelper: AzureAsrHelper
- private var ttsHelper: AzureTtsHelper
- private static var eventStreamHandler: AzureEventStreamHandler?
-
- // 创建方法到通道的映射
- private static var ttsMethodHandlers = [String: FlutterMethodCallHandler]()
- private static var asrMethodHandlers = [String: FlutterMethodCallHandler]()
-
- public static func register(with registrar: FlutterPluginRegistrar) {
- // ASR通道
- let channel = FlutterMethodChannel(name: "com.deep_voice.azure_asr", binaryMessenger: registrar.messenger())
-
- // TTS通道
- let ttsChannel = FlutterMethodChannel(name: "com.deep_voice.azure_tts", binaryMessenger: registrar.messenger())
-
- // 设置ASR事件通道
- let eventChannel = FlutterEventChannel(name: "com.deep_voice.azure_asr_events", binaryMessenger: registrar.messenger())
- eventStreamHandler = AzureEventStreamHandler()
- eventChannel.setStreamHandler(eventStreamHandler)
-
- let instance = SwiftAzureSpeechRecognitionPlugin(
- azureChannel: channel,
- ttsChannel: ttsChannel,
- eventStreamHandler: eventStreamHandler!
- )
-
- // 直接设置各自通道的处理器
- channel.setMethodCallHandler(instance.handleAsrMethodCalls)
- ttsChannel.setMethodCallHandler(instance.handleTtsMethodCalls)
- }
-
-
-
- // 新增直接处理方法调用的函数
- private func handleTtsMethodCalls(_ call: FlutterMethodCall, result: @escaping FlutterResult) {
- handleTtsMethod(call, result)
- }
-
- private func handleAsrMethodCalls(_ call: FlutterMethodCall, result: @escaping FlutterResult) {
- handleAsrMethod(call, result)
- }
-
-
- init(azureChannel: FlutterMethodChannel, ttsChannel: FlutterMethodChannel, eventStreamHandler: AzureEventStreamHandler) {
- self.azureChannel = azureChannel
- self.ttsChannel = ttsChannel
-
- // 创建辅助类实例，使用自定义事件回调处理器
- let eventHandler: (String, [String: Any]) -> Void = { eventName, arguments in
- DispatchQueue.main.async {
- if let eventSink = SwiftAzureSpeechRecognitionPlugin.eventStreamHandler?.eventSink {
- var eventData = arguments
- eventData["type"] = eventName
- eventSink(eventData)
- }
- }
- }
-
- asrHelper = AzureAsrHelper(eventHandler: eventHandler)
- ttsHelper = AzureTtsHelper(eventHandler: eventHandler)
-
- super.init()
- }
-
- private func handleAsrMethod(_ call: FlutterMethodCall, _ result: @escaping FlutterResult) {
-
- let args = call.arguments as? Dictionary
-
- switch call.method {
- case "initialize":
- // 仅在初始化时读取必要参数
- guard let speechSubscriptionKey = args?["subscriptionKey"] as? String, !speechSubscriptionKey.isEmpty else {
- let errorMsg = "语音订阅密钥不能为空"
- print("[AzurePlugin] 错误: \(errorMsg)")
- result(FlutterError(code: "INVALID_SUBSCRIPTION_KEY", message: errorMsg, details: nil))
- return
- }
-
- guard let serviceRegion = args?["region"] as? String, !serviceRegion.isEmpty else {
- let errorMsg = "服务区域不能为空"
- print("[AzurePlugin] 错误: \(errorMsg)")
- result(FlutterError(code: "INVALID_REGION", message: errorMsg, details: nil))
- return
- }
-
- let supportedLanguages = args?["supportedLanguages"] as? [String] ?? []
-
- let success = asrHelper.initialize(
- speechSubscriptionKey: speechSubscriptionKey,
- serviceRegion: serviceRegion,
- supportedLanguages: supportedLanguages.isEmpty ? nil : supportedLanguages
- )
- result(success)
-
- case "startContinuousRecognition":
- // 只有使用参数时才验证
- let success = asrHelper.startContinuousRecognition()
- result(success)
-
- case "stopContinuousRecognition":
- // 不需要额外参数
- let success = asrHelper.stopContinuousRecognition()
- result(success)
-
- case "recognizeOnce":
- // 只有使用参数时才验证
- let success = asrHelper.recognizeOnce()
- result(success)
-
- case "isContinuousRecognitionActive":
- // 不需要额外参数
- result(asrHelper.isContinuousRecognitionActive())
-
- case "dispose":
- // 不需要额外参数
- print("[AzurePlugin] 释放ASR资源")
- asrHelper.dispose()
- result(true)
-
- default:
- print("[AzurePlugin] 错误: 未知ASR方法: \(call.method)")
- result(FlutterMethodNotImplemented)
- }
- }
-
- private func handleTtsMethod(_ call: FlutterMethodCall, _ result: @escaping FlutterResult) {
-
- let args = call.arguments as? Dictionary
-
- switch call.method {
- case "initialize":
- // 仅在初始化时验证参数
- guard let speechSubscriptionKey = args?["subscriptionKey"] as? String, !speechSubscriptionKey.isEmpty else {
- let errorMsg = "语音订阅密钥不能为空"
- print("[AzurePlugin] 错误: \(errorMsg)")
- result(FlutterError(code: "INVALID_SUBSCRIPTION_KEY", message: errorMsg, details: nil))
- return
- }
-
- guard let serviceRegion = args?["region"] as? String, !serviceRegion.isEmpty else {
- let errorMsg = "服务区域不能为空"
- print("[AzurePlugin] 错误: \(errorMsg)")
- result(FlutterError(code: "INVALID_REGION", message: errorMsg, details: nil))
- return
- }
-
- let language = args?["language"] as? String ?? "zh-CN"
-
- print("[AzurePlugin] 初始化TTS，语言: \(language)")
-
- let success = ttsHelper.initialize(speechSubscriptionKey: speechSubscriptionKey, serviceRegion: serviceRegion, language: language)
- result(success)
-
- case "setVoice":
- // 仅获取voice参数
- guard let voiceName = args?["voiceName"] as? String, !voiceName.isEmpty else {
- let errorMsg = "声音名称不能为空"
- print("[AzurePlugin] 错误: \(errorMsg)")
- result(FlutterError(code: "INVALID_VOICE", message: errorMsg, details: nil))
- return
- }
-
- print("[AzurePlugin] 设置声音: \(voiceName)")
-
- let success = ttsHelper.setVoice(voiceName: voiceName)
- result(success)
-
- case "speakText":
- // 仅获取text参数
- let text = args?["text"] as? String ?? ""
-
- if text.isEmpty {
- print("[AzurePlugin] 警告: 要播放的文本为空")
- result("OK")
- return
- }
-
- print("[AzurePlugin] 播放文本: \(text.prefix(50))...")
-
- let success = ttsHelper.speakText(text: text)
- result(success ? "OK" : "ERROR")
-
- case "speakSsml":
- // 仅获取ssml参数
- guard let ssml = args?["ssml"] as? String, !ssml.isEmpty else {
- let errorMsg = "SSML内容不能为空"
- print("[AzurePlugin] 错误: \(errorMsg)")
- result(FlutterError(code: "INVALID_SSML", message: errorMsg, details: nil))
- return
- }
-
- print("[AzurePlugin] 播放SSML: \(ssml.prefix(100))...")
-
- // 由于我们移除了speakSsml方法，这里改用speakText方法
- // Azure SDK内部会自动检测是普通文本还是SSML
- let success = ttsHelper.speakText(text: ssml)
- result(success)
-
- case "stopSpeaking":
- // 不需要参数
- print("[AzurePlugin] 停止播放")
- let success = ttsHelper.stopSpeaking()
- result(success)
-
- case "isSpeaking":
- // 不需要参数
- result(ttsHelper.isSpeaking())
-
- case "setSpeechParams":
- // 仅获取语音参数
- let rate = args?["rate"] as? Int ?? 0
- let pitch = args?["pitch"] as? Int ?? 0
- let volume = args?["volume"] as? Int ?? 100
-
- print("[AzurePlugin] 设置语音参数: rate=\(rate), pitch=\(pitch), volume=\(volume)")
- let success = ttsHelper.setSpeechParams(rate: rate, pitch: pitch, volume: volume)
- result(success)
-
- case "dispose":
- // 释放TTS资源
- print("[AzurePlugin] 释放TTS资源")
- ttsHelper.dispose()
- result(true)
-
- default:
- print("[AzurePlugin] 错误: 未知TTS方法: \(call.method)")
- result(FlutterMethodNotImplemented)
- }
- }
-}
-
-// 用于处理事件流的辅助类
-@available(iOS 13.0, *)
-class AzureEventStreamHandler: NSObject, FlutterStreamHandler {
- var eventSink: FlutterEventSink?
-
- func onListen(withArguments arguments: Any?, eventSink events: @escaping FlutterEventSink) -> FlutterError? {
- self.eventSink = events
- // 通知Flutter端事件通道已准备好
- DispatchQueue.main.async {
- events(["type": "channelReady"])
- }
- return nil
- }
-
- func onCancel(withArguments arguments: Any?) -> FlutterError? {
- self.eventSink = nil
- return nil
- }
-}
\ No newline at end of file
diff --git a/azure/ios/azure_speech_recognition.podspec b/azure/ios/azure_speech_recognition.podspec
deleted file mode 100644
index 88797fb0c..000000000
--- a/azure/ios/azure_speech_recognition.podspec
+++ /dev/null
@@ -1,24 +0,0 @@
-#
-# To learn more about a Podspec see http://guides.cocoapods.org/syntax/podspec.html.
-# Run `pod lib lint azure_speech_recognition.podspec` to validate before publishing.
-#
-Pod::Spec.new do |s|
- s.name = 'azure_speech_recognition'
- s.version = '0.1.0'
- s.summary = 'Azure Speech Recognition plugin for Flutter'
- s.description = <<-DESC
-A Flutter plugin for Microsoft Azure Speech services, providing both speech recognition (ASR) and text-to-speech (TTS) capabilities.
- DESC
- s.homepage = 'https://github.com/yourusername/azure_speech_recognition'
- s.license = { :type => 'MIT', :file => '../LICENSE' }
- s.author = { 'Your Company' => 'your-email@example.com' }
- s.source = { :path => '.' }
- s.source_files = 'Classes/**/*'
- s.dependency 'Flutter'
- s.dependency 'MicrosoftCognitiveServicesSpeech-iOS', '~> 1.34.0'
- s.platform = :ios, '12.0'
-
- # Flutter.framework does not contain a i386 slice.
- s.pod_target_xcconfig = { 'DEFINES_MODULE' => 'YES', 'EXCLUDED_ARCHS[sdk=iphonesimulator*]' => 'i386' }
- s.swift_version = '5.0'
-end
\ No newline at end of file
diff --git a/azure/lib/azure_speech_recognition.dart b/azure/lib/azure_speech_recognition.dart
deleted file mode 100644
index 02b3f92d8..000000000
--- a/azure/lib/azure_speech_recognition.dart
+++ /dev/null
@@ -1,6 +0,0 @@
-// This is a placeholder file that exports nothing.
-// The actual implementation is in the app's services folder.
-// This file exists just to satisfy the Flutter plugin structure requirements.
-
-// Empty library to satisfy plugin structure
-library azure_speech_recognition;
\ No newline at end of file
diff --git a/azure/pubspec.yaml b/azure/pubspec.yaml
deleted file mode 100644
index a36c77bdc..000000000
--- a/azure/pubspec.yaml
+++ /dev/null
@@ -1,23 +0,0 @@
-name: azure_speech_recognition
-description: Azure Speech Recognition and Text-to-Speech services Flutter plugin
-version: 0.1.0
-homepage: https://github.com/yourusername/azure_speech_recognition
-
-environment:
- sdk: '>=2.12.0 <3.0.0'
- flutter: ">=2.0.0"
-
-dependencies:
- flutter:
- sdk: flutter
-
-dev_dependencies:
- flutter_test:
- sdk: flutter
- flutter_lints: ^1.0.0
-
-flutter:
- plugin:
- platforms:
- ios:
- pluginClass: AzureSpeechRecognitionPlugin
\ No newline at end of file
diff --git a/lib/core/bindings/initial_binding.dart b/lib/core/bindings/initial_binding.dart
index 3f42d9560..748589f63 100644
--- a/lib/core/bindings/initial_binding.dart
+++ b/lib/core/bindings/initial_binding.dart
@@ -8,6 +8,7 @@ import '../../data/services/classic_bluetooth_service.dart';
import '../../data/services/bluetooth_media_button_service.dart';
import '../../core/utils/logger.dart';
import '../../data/services/voice_interaction_service.dart';
+import '../../data/services/open_ai_service_adapter.dart';
/// 初始绑定，用于管理全局依赖
class InitialBinding extends Bindings {
@@ -29,6 +30,13 @@ class InitialBinding extends Bindings {
Get.lazyPut(() => VolcanoTranslationService(),
fenix: true);
+ // OpenAI服务适配器
+ Get.lazyPut(() {
+ final adapter = OpenAIServiceAdapter();
+ adapter.initialize();
+ return adapter;
+ }, fenix: true);
+
// 火山AI服务
Get.lazyPut(() => VolcanoAIService(), fenix: true);
@@ -41,6 +49,7 @@ class InitialBinding extends Bindings {
() => BluetoothMediaButtonService(),
fenix: true);
+ // 语音交互服务
Get.lazyPut(() => VoiceInteractionService(),
fenix: true);
diff --git a/lib/data/models/events/voice_interaction_event.dart b/lib/data/models/events/voice_interaction_event.dart
index 4bfa636df..81860de23 100644
--- a/lib/data/models/events/voice_interaction_event.dart
+++ b/lib/data/models/events/voice_interaction_event.dart
@@ -26,4 +26,17 @@ class RecognitionStartedEvent extends VoiceInteractionEvent {
RecognitionStartedEvent({
required int timestamp,
}) : super(timestamp: timestamp);
+}
+
+/// 通用语音交互事件
+/// 用于处理其他类型的事件
+class GenericVoiceInteractionEvent extends VoiceInteractionEvent {
+ final String type;
+ final Map? data;
+
+ GenericVoiceInteractionEvent({
+ required this.type,
+ this.data,
+ required int timestamp,
+ }) : super(timestamp: timestamp);
}
\ No newline at end of file
diff --git a/lib/data/services/ai_service.dart b/lib/data/services/ai_service.dart
index 949dfca67..a2232b9b5 100644
--- a/lib/data/services/ai_service.dart
+++ b/lib/data/services/ai_service.dart
@@ -1,5 +1,7 @@
/// AI回复服务接口
abstract class AiService {
+
+
/// 非流式输出方法
Future sendMessage({
required List> messages,
diff --git a/lib/data/services/open_ai_service_adapter.dart b/lib/data/services/open_ai_service_adapter.dart
new file mode 100644
index 000000000..2c0da53ac
--- /dev/null
+++ b/lib/data/services/open_ai_service_adapter.dart
@@ -0,0 +1,247 @@
+import 'dart:async';
+import 'package:flutter_dotenv/flutter_dotenv.dart';
+import 'package:open_ai_service/open_ai_service.dart';
+import 'package:get/get.dart';
+import 'ai_service.dart';
+
+/// OpenAI服务适配器 - 连接AiService接口与OpenAIService插件
+class OpenAIServiceAdapter implements AiService {
+ final OpenAIService _openAIService = OpenAIService();
+ StreamSubscription? _eventSubscription;
+ final StreamController _tokenStreamController = StreamController.broadcast();
+ bool _isProcessingStream = false;
+
+ /// 构造函数
+ OpenAIServiceAdapter() {
+ printInfo(info: '创建OpenAIServiceAdapter实例');
+ _setupEventListener();
+ }
+
+ /// 设置事件监听器
+ void _setupEventListener() {
+ try {
+ // 首先确保访问eventStream以初始化底层事件通道
+ _openAIService.eventStream;
+
+ // 设置事件处理
+ _eventSubscription = _openAIService.eventStream.listen(
+ (event) {
+ if (!_isProcessingStream) return;
+
+ try {
+ switch (event.type) {
+ case OpenAIEventType.token:
+ if (event.content is String) {
+ _tokenStreamController.add(event.content as String);
+ } else {
+ printInfo(info: '收到非字符串类型的token: ${event.content}');
+ }
+ break;
+ case OpenAIEventType.complete:
+ _isProcessingStream = false;
+ break;
+ case OpenAIEventType.error:
+ if (event.content is String) {
+ _tokenStreamController.addError(event.content as String);
+ } else {
+ _tokenStreamController.addError('未知错误: ${event.content}');
+ }
+ _isProcessingStream = false;
+ break;
+ case OpenAIEventType.functionCall:
+ try {
+ if (event.content is Map) {
+ _tokenStreamController.addError('收到函数调用，该流仅支持文本响应');
+ } else {
+ _tokenStreamController.addError('收到未知格式的函数调用');
+ printError(info: '函数调用格式错误: ${event.content}');
+ }
+ } catch (e) {
+ printError(info: '处理函数调用事件出错: $e');
+ _tokenStreamController.addError('处理函数调用失败: $e');
+ }
+ _isProcessingStream = false;
+ break;
+ }
+ } catch (e) {
+ printError(info: '处理事件出错: $e');
+ _tokenStreamController.addError('处理事件失败: $e');
+ _isProcessingStream = false;
+ }
+ },
+ onError: (error) {
+ printError(info: '事件流错误: $error');
+ _tokenStreamController.addError('事件流错误: $error');
+ _isProcessingStream = false;
+ },
+ onDone: () {
+ printInfo(info: '事件流已关闭');
+ _isProcessingStream = false;
+ },
+ );
+ } catch (e) {
+ printError(info: '设置事件监听器失败: $e');
+ }
+ }
+
+ /// 初始化OpenAI服务
+ Future initialize() async {
+ try {
+ // 从.env文件中读取配置
+ final apiKey = dotenv.env['OPENAI_API_KEY'] ?? '';
+ final baseUrl = dotenv.env['OPENAI_BASE_URL'] ?? '';
+ final model = dotenv.env['OPENAI_MODEL'] ?? '';
+
+ printInfo(info: '从.env读取OpenAI配置');
+ printInfo(info: '基础URL: $baseUrl');
+ printInfo(info: '模型名称: $model');
+
+ if (apiKey.isEmpty) {
+ printError(info: '错误: OpenAI API密钥未配置，请在.env文件中设置OPENAI_API_KEY');
+ return false;
+ }
+
+ // 初始化OpenAI服务
+ final result = await _openAIService.initialize(
+ apiKey: apiKey,
+ baseUrl: baseUrl,
+ model: model,
+ );
+
+ if (result) {
+ printInfo(info: 'OpenAI服务初始化成功');
+ } else {
+ printError(info: 'OpenAI服务初始化失败');
+ }
+
+ return result;
+ } catch (e) {
+ printError(info: 'OpenAI服务初始化异常: $e');
+ return false;
+ }
+ }
+
+ /// 发送消息并获取回复
+ @override
+ Future sendMessage({
+ required List> messages,
+ required String systemPrompt,
+ }) async {
+ try {
+ // 在方法内直接转换
+ final convertedMessages = messages.map((m) =>
+ Map.from(m)).toList();
+
+ // 发送消息并获取回复
+ final response = await _openAIService.sendMessage(
+ messages: convertedMessages,
+ );
+
+ return response;
+ } catch (e) {
+ printError(info: 'OpenAI发送消息失败: $e');
+ throw '发送消息失败: $e';
+ }
+ }
+
+ /// 发送消息并获取流式回复
+ @override
+ Stream sendMessageStream({
+ required List> messages,
+ required String systemPrompt,
+ }) async* {
+ try {
+ // 在方法内直接转换
+ final convertedMessages = messages.map((m) =>
+ Map.from(m)).toList();
+
+ // 创建用于接收token的控制器
+ final localController = StreamController();
+
+ // 标记开始处理流
+ _isProcessingStream = true;
+
+ // 添加从广播流到本地流的订阅
+ final subscription = _tokenStreamController.stream.listen(
+ (token) => localController.add(token),
+ onError: (error) {
+ printError(info: '令牌流错误: $error');
+ localController.addError(error);
+ localController.close();
+ },
+ onDone: () {
+ printInfo(info: '令牌流已完成');
+ localController.close();
+ }
+ );
+
+ // 当本地控制器关闭时，取消订阅
+ localController.onCancel = () {
+ subscription.cancel();
+ };
+
+ // 启动流式消息请求
+ bool started = false;
+ try {
+ started = await _openAIService.sendMessageStream(
+ messages: convertedMessages,
+ );
+ } catch (e) {
+ printError(info: '启动消息流失败: $e');
+ localController.addError('启动消息流失败: $e');
+ localController.close();
+ _isProcessingStream = false;
+ throw '启动消息流失败: $e';
+ }
+
+ if (!started) {
+ printError(info: '无法启动消息流');
+ localController.addError('无法启动消息流');
+ localController.close();
+ _isProcessingStream = false;
+ throw '无法启动消息流';
+ }
+
+ // 通过yield*将controller的流转发
+ yield* localController.stream;
+ } catch (e) {
+ printError(info: 'OpenAI流式消息处理失败: $e');
+ throw '流式消息处理失败: $e';
+ }
+ }
+
+ /// 注册函数
+ Future registerFunction(String name, String description, Map parameters) async {
+ try {
+ final result = await _openAIService.registerFunction(
+ name: name,
+ description: description,
+ parameters: parameters,
+ );
+
+ if (result) {
+ printInfo(info: '函数 "$name" 注册成功');
+ } else {
+ printError(info: '函数 "$name" 注册失败');
+ }
+
+ return result;
+ } catch (e) {
+ printError(info: '注册函数失败: $e');
+ return false;
+ }
+ }
+
+ /// 释放资源
+ void dispose() {
+ try {
+ _isProcessingStream = false;
+ _eventSubscription?.cancel();
+ _tokenStreamController.close();
+ printInfo(info: 'OpenAIServiceAdapter资源已释放');
+ } catch (e) {
+ printError(info: '释放资源时出错: $e');
+ }
+ }
+
+}
\ No newline at end of file
diff --git a/lib/data/services/speech_impl/azure_asr_service.dart b/lib/data/services/speech_impl/azure_asr_service.dart
index ffb67ed46..e09defb66 100644
--- a/lib/data/services/speech_impl/azure_asr_service.dart
+++ b/lib/data/services/speech_impl/azure_asr_service.dart
@@ -10,8 +10,8 @@ import '../asr_service.dart';
/// 该服务提供了通过平台通道与 Android 上的 Microsoft Speech SDK 交互的接口
class AzureAsrService extends GetxService implements AsrService {
static final AzureAsrService to = Get.put(AzureAsrService());
- static const MethodChannel _channel = MethodChannel('com.deep_voice.azure_asr');
- static const EventChannel _eventChannel = EventChannel('com.deep_voice.azure_asr_events');
+ static const MethodChannel _channel = MethodChannel('azure_speech/asr');
+ static const EventChannel _eventChannel = EventChannel('azure_speech/asr_events');
bool _isInitialized = false;
late final String _subscriptionKey;
diff --git a/lib/data/services/speech_impl/azure_tts_service.dart b/lib/data/services/speech_impl/azure_tts_service.dart
index ae5ab9369..1cdb62507 100644
--- a/lib/data/services/speech_impl/azure_tts_service.dart
+++ b/lib/data/services/speech_impl/azure_tts_service.dart
@@ -12,7 +12,7 @@ import '../tts_service.dart';
/// 提供文本转语音功能。
class AzureTtsService extends GetxService implements TtsService {
static final AzureTtsService to = Get.put(AzureTtsService());
- static const MethodChannel _channel = MethodChannel('com.deep_voice.azure_tts');
+ static const MethodChannel _channel = MethodChannel('azure_speech/tts');
bool _isInitialized = false;
late final String _subscriptionKey;
diff --git a/lib/data/services/voice_interaction_service.dart b/lib/data/services/voice_interaction_service.dart
index a1d26d509..cce506a58 100644
--- a/lib/data/services/voice_interaction_service.dart
+++ b/lib/data/services/voice_interaction_service.dart
@@ -5,6 +5,7 @@ import 'package:flutter_dotenv/flutter_dotenv.dart';
import '../models/events/voice_interaction_event.dart';
import '../../core/utils/logger.dart';
import '../../modules/chat/models/message_model.dart';
+import '../../routes/app_routes.dart';
import 'chat_history_service.dart';
/// 语音交互服务接口
@@ -38,8 +39,9 @@ class VoiceInteractionService extends GetxService {
// 配置信息
late String _azureSpeechKey;
late String _azureSpeechRegion;
- late String _volcanoAiKey;
-
+ late String _openaiApiKey;
+ late String _openaiBaseUrl;
+ late String _openaiModel;
// 聊天历史服务
late final ChatHistoryService _chatHistoryService;
@@ -57,15 +59,14 @@ class VoiceInteractionService extends GetxService {
void _loadConfig() {
_azureSpeechKey = dotenv.env['AZURE_SPEECH_KEY'] ?? '';
_azureSpeechRegion = dotenv.env['AZURE_SPEECH_REGION'] ?? '';
- _volcanoAiKey = dotenv.env['VOLCANO_AI_API_KEY'] ?? '';
+ _openaiApiKey = dotenv.env['OPENAI_API_KEY'] ?? '';
+ _openaiBaseUrl = dotenv.env['OPENAI_BASE_URL'] ?? '';
+ _openaiModel = dotenv.env['OPENAI_MODEL'] ?? '';
if (_azureSpeechKey.isEmpty || _azureSpeechRegion.isEmpty) {
Logger.warning('未找到 Azure 语音服务配置。请在 .env 文件中设置 AZURE_SPEECH_KEY 和 AZURE_SPEECH_REGION');
}
-
- if (_volcanoAiKey.isEmpty) {
- Logger.warning('未找到火山 AI API 密钥。请在 .env 文件中设置 VOLCANO_AI_API_KEY');
- }
+
}
/// 处理来自原生层的事件
@@ -104,6 +105,29 @@ class VoiceInteractionService extends GetxService {
// 保存聊天历史到ChatHistoryService
_saveChatHistory(agentId, userMessage, assistantMessage, DateTime.now().millisecondsSinceEpoch);
break;
+
+ case 'enter_translation_mode':
+ // 进入翻译模式事件
+ Logger.info('收到进入翻译模式事件，正在导航到翻译界面');
+ _navigateToTranslation();
+
+ final translationModeEvent = GenericVoiceInteractionEvent(
+ type: 'enter_translation_mode',
+ timestamp: DateTime.now().millisecondsSinceEpoch,
+ );
+ _eventStreamController.add(translationModeEvent);
+ break;
+ }
+ }
+
+ /// 导航到翻译界面
+ void _navigateToTranslation() {
+ try {
+ // 使用GetX导航到翻译页面
+ Get.toNamed(Routes.translation);
+ Logger.info('已导航到翻译界面');
+ } catch (e) {
+ Logger.error('导航到翻译界面失败: $e');
}
}
@@ -195,7 +219,9 @@ class VoiceInteractionService extends GetxService {
final result = await _channel.invokeMethod('startService', {
'azure_speech_key': _azureSpeechKey,
'azure_speech_region': _azureSpeechRegion,
- 'volcano_ai_api_key': _volcanoAiKey,
+ 'openai_api_key': _openaiApiKey,
+ 'openai_base_url': _openaiBaseUrl,
+ 'openai_model': _openaiModel,
}) ?? false;
if (result) {
diff --git a/lib/modules/chat/controllers/chat_controller.dart b/lib/modules/chat/controllers/chat_controller.dart
index f882e342f..8f446444e 100644
--- a/lib/modules/chat/controllers/chat_controller.dart
+++ b/lib/modules/chat/controllers/chat_controller.dart
@@ -17,6 +17,7 @@ import '../../../data/services/asr_service.dart';
import '../../../data/services/chat_history_service.dart';
import '../../../data/models/events/voice_interaction_event.dart';
import '../../../data/services/voice_interaction_service.dart';
+import '../../../data/services/open_ai_service_adapter.dart';
class ChatController extends GetxController {
// 服务
@@ -97,6 +98,7 @@ class ChatController extends GetxController {
agent = foundAgent;
+ // 根据Agent ID选择不同的AI服务
switch (agent.id) {
case 'cyber_girlfriend':
// 亲子陪伴
@@ -109,7 +111,15 @@ class ChatController extends GetxController {
_aiService = Get.find();
break;
default:
- _aiService = Get.find();
+ // 默认使用OpenAIServiceAdapter
+ try {
+ _aiService = Get.find();
+ Logger.info('使用OpenAIServiceAdapter');
+ } catch (e) {
+ // 如果找不到OpenAIServiceAdapter，则回退到VolcanoAIService
+ Logger.info('未找到OpenAIServiceAdapter，回退使用VolcanoAIService: $e');
+ _aiService = Get.find();
+ }
}
// 使用克隆音色语音合成
diff --git a/local_plugins/azure_speech/README.md b/local_plugins/azure_speech/README.md
new file mode 100644
index 000000000..067b344eb
--- /dev/null
+++ b/local_plugins/azure_speech/README.md
@@ -0,0 +1,136 @@
+# Azure Speech 插件
+
+本插件为Flutter提供了Azure语音服务的集成，包括：
+
+- 语音合成（TTS）
+- 语音识别（ASR）
+
+## 功能
+
+### 语音合成（TTS）
+
+- 支持多种语音（如中文、英文等）
+- 语音参数调整（语速、音调、音量）
+- 音频输出设备选择（扬声器、听筒、自动）
+- SSML支持
+
+### 语音识别（ASR）
+
+- 一次性语音识别
+- 连续语音识别
+- 自动语言检测
+- 音频处理优化（回音消除、噪声抑制等）
+
+## 平台支持
+
+- Android
+- iOS
+
+## 如何使用
+
+### 初始化
+
+```dart
+import 'package:azure_speech/azure_speech.dart';
+
+// 初始化TTS
+await AzureSpeech.initializeTts(
+ 'your_subscription_key',
+ 'your_service_region',
+ language: 'zh-CN',
+);
+
+// 初始化ASR
+await AzureSpeech.initializeAsr(
+ 'your_subscription_key',
+ 'your_service_region',
+ ['zh-CN', 'en-US'],
+);
+```
+
+### 语音合成
+
+```dart
+// 设置语音
+await AzureSpeech.setTtsVoice('zh-CN-XiaoxiaoNeural');
+
+// 设置语音参数
+await AzureSpeech.setTtsSpeechParams(
+ rate: 0, // 语速 -100~100
+ pitch: 0, // 音调 -100~100
+ volume: 100, // 音量 0~100
+);
+
+// 设置音频输出设备
+await AzureSpeech.setTtsAudioOutputType('AUTO'); // 'SPEAKER', 'EARPIECE', 'AUTO'
+
+// 播放文本
+await AzureSpeech.speakText('你好，世界！');
+
+// 停止播放
+await AzureSpeech.stopSpeaking();
+
+// 检查是否正在播放
+bool isSpeaking = await AzureSpeech.isSpeaking();
+```
+
+### 语音识别
+
+```dart
+// 一次性识别
+final result = await AzureSpeech.recognizeOnce();
+if (result['success']) {
+ print('识别文本: ${result['text']}');
+ print('识别语言: ${result['language']}');
+} else {
+ print('识别失败: ${result['error']}');
+}
+
+// 连续识别
+// 监听识别结果
+AzureSpeech.asrResultStream.listen((event) {
+ switch (event['eventType']) {
+ case 'recognizing':
+ // 实时识别中的结果
+ print('识别中: ${event['text']}');
+ break;
+ case 'finalResult':
+ // 最终识别结果
+ print('最终结果: ${event['text']}');
+ break;
+ case 'error':
+ // 错误
+ print('错误: ${event['error']}');
+ break;
+ }
+});
+
+// 开始连续识别
+await AzureSpeech.startContinuousRecognition();
+
+// 停止连续识别
+await AzureSpeech.stopContinuousRecognition();
+
+// 检查连续识别是否活跃
+bool isActive = await AzureSpeech.isContinuousRecognitionActive();
+```
+
+### 释放资源
+
+```dart
+// 释放所有资源
+await AzureSpeech.dispose();
+```
+
+## 依赖项
+
+本插件依赖于：
+
+- Microsoft Cognitive Services Speech SDK
+- Flutter
+
+## 注意事项
+
+- 使用前需要在Azure门户中创建语音服务资源，并获取订阅密钥和区域
+- Android需要相关权限：RECORD_AUDIO, INTERNET等
+- iOS需要在Info.plist中添加麦克风使用权限描述
\ No newline at end of file
diff --git a/local_plugins/azure_speech/android/build.gradle.kts b/local_plugins/azure_speech/android/build.gradle.kts
new file mode 100644
index 000000000..a26b13967
--- /dev/null
+++ b/local_plugins/azure_speech/android/build.gradle.kts
@@ -0,0 +1,64 @@
+import com.android.build.gradle.LibraryExtension
+
+buildscript {
+ repositories {
+ google()
+ mavenCentral()
+ }
+ dependencies {
+ classpath("com.android.tools.build:gradle:7.3.0")
+ classpath("org.jetbrains.kotlin:kotlin-gradle-plugin:1.7.10")
+ }
+}
+
+allprojects {
+ repositories {
+ google()
+ mavenCentral()
+ }
+}
+
+plugins {
+ id("com.android.library")
+ kotlin("android")
+}
+
+// 配置android扩展
+configure {
+ namespace = "com.yunqiinnovation.azure_speech"
+ compileSdkVersion(33)
+
+ defaultConfig {
+ minSdk = 21
+ }
+
+ compileOptions {
+ sourceCompatibility = JavaVersion.VERSION_11
+ targetCompatibility = JavaVersion.VERSION_11
+ }
+
+ sourceSets {
+ getByName("main") {
+ manifest.srcFile("src/main/AndroidManifest.xml")
+ java.srcDirs("src/main/kotlin")
+ }
+ }
+
+ // 添加lint选项
+ lintOptions {
+ isCheckReleaseBuilds = false
+ }
+}
+
+// 显式设置Kotlin JVM目标版本
+tasks.withType {
+ kotlinOptions {
+ jvmTarget = "11"
+ }
+}
+
+dependencies {
+
+ // 添加Microsoft语音SDK
+ implementation("com.microsoft.cognitiveservices.speech:client-sdk:1.30.0")
+}
\ No newline at end of file
diff --git a/local_plugins/azure_speech/android/settings.gradle.kts b/local_plugins/azure_speech/android/settings.gradle.kts
new file mode 100644
index 000000000..14ba24107
--- /dev/null
+++ b/local_plugins/azure_speech/android/settings.gradle.kts
@@ -0,0 +1 @@
+rootProject.name = "azure_speech"
diff --git a/local_plugins/azure_speech/android/src/main/AndroidManifest.xml b/local_plugins/azure_speech/android/src/main/AndroidManifest.xml
new file mode 100644
index 000000000..df2a770c6
--- /dev/null
+++ b/local_plugins/azure_speech/android/src/main/AndroidManifest.xml
@@ -0,0 +1,6 @@
+
+
+
+
+
\ No newline at end of file
diff --git a/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureAsrHelper.kt b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureAsrHelper.kt
new file mode 100644
index 000000000..1dd7cf624
--- /dev/null
+++ b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureAsrHelper.kt
@@ -0,0 +1,593 @@
+package com.yunqiinnovation.azure_speech
+
+import android.content.Context
+import android.media.AudioAttributes
+import android.media.AudioFormat
+import android.media.AudioRecord
+import android.media.MediaRecorder
+import android.media.audiofx.AcousticEchoCanceler
+import android.media.audiofx.NoiseSuppressor
+import android.media.audiofx.AutomaticGainControl
+import android.os.Process
+import com.yunqiinnovation.azure_speech.utils.FileLogger
+import com.microsoft.cognitiveservices.speech.*
+import com.microsoft.cognitiveservices.speech.audio.*
+import com.microsoft.cognitiveservices.speech.util.EventHandler
+import java.util.concurrent.ExecutionException
+import java.util.concurrent.atomic.AtomicBoolean
+
+class AzureAsrHelper(private val context: Context) {
+ private var recognizer: SpeechRecognizer? = null
+ private var speechConfig: SpeechConfig? = null
+ private val TAG = "AzureAsrHelper"
+ private var isContinuousRecognitionActive = false
+ private var currentLanguage = "zh-CN"
+ private var subscriptionKey = ""
+ private var region = ""
+ private var isAutoDetectLanguage = false
+ private var supportedLanguages = arrayOf("zh-CN")
+
+ // 是否使用回音消除 - 内部控制常量
+ private val useEchoCancellation = false
+
+ // 自定义音频处理相关
+ private var customAudioProcessor: CustomAudioProcessor? = null
+ private var pushStream: PushAudioInputStream? = null
+ private var audioConfig: AudioConfig? = null
+
+ // 初始化SDK并创建recognizer
+ fun initialize(subscriptionKey: String, region: String,
+ supportedLanguages: Array = arrayOf("zh-CN")): Boolean {
+ try {
+ FileLogger.d(TAG, "初始化 Azure 语音服务")
+
+ // 检查配置是否为空
+ if (subscriptionKey.isEmpty() || region.isEmpty()) {
+ FileLogger.e(TAG, "Azure 配置信息不完整")
+ return false
+ }
+
+ // 释放之前的资源
+ dispose()
+
+ this.subscriptionKey = subscriptionKey
+ this.region = region
+
+ // 设置语言
+ if (supportedLanguages.isNotEmpty()) {
+ this.supportedLanguages = supportedLanguages
+ }
+
+ // 根据支持的语言数量决定是否启用自动语言检测
+ this.isAutoDetectLanguage = supportedLanguages.size >= 2
+
+ // 如果只有一种语言，设置为当前语言
+ if (!isAutoDetectLanguage && supportedLanguages.isNotEmpty()) {
+ this.currentLanguage = supportedLanguages[0]
+ }
+
+ // 创建语音配置
+ speechConfig = SpeechConfig.fromSubscription(subscriptionKey, region)
+
+ // 设置语言配置
+ if (isAutoDetectLanguage) {
+ // 设置自动语言检测
+ speechConfig?.setProperty(PropertyId.SpeechServiceConnection_LanguageIdMode, "Continuous")
+ } else {
+ // 设置指定的识别语言
+ speechConfig?.speechRecognitionLanguage = currentLanguage
+ }
+
+ // 创建识别器
+ try {
+ if (useEchoCancellation) {
+ // 如果使用回音消除，创建自定义音频输入流
+ setupCustomAudioProcessing()
+
+ if (isAutoDetectLanguage) {
+ val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
+ recognizer = SpeechRecognizer(speechConfig, autoDetectConfig, audioConfig)
+ } else {
+ recognizer = SpeechRecognizer(speechConfig, audioConfig)
+ }
+ } else {
+ // 使用默认麦克风输入
+ if (isAutoDetectLanguage) {
+ val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
+ recognizer = SpeechRecognizer(speechConfig, autoDetectConfig)
+ } else {
+ recognizer = SpeechRecognizer(speechConfig)
+ }
+ }
+
+ FileLogger.d(TAG, "Azure 语音服务初始化成功")
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "创建识别器失败: ${e.message}")
+ stopCustomAudioProcessing()
+ return false
+ }
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "初始化失败: ${e.message}")
+ return false
+ }
+ }
+
+ // 重置 recognizer
+ private fun resetRecognizer(): Boolean {
+ try {
+ // 释放之前的 recognizer
+ recognizer?.close()
+ recognizer = null
+
+ // 停止当前的音频处理
+ stopCustomAudioProcessing()
+
+ // 使用现有配置重新创建 recognizer
+ if (speechConfig != null) {
+ if (useEchoCancellation) {
+ // 如果使用回音消除，创建自定义音频输入流
+ setupCustomAudioProcessing()
+
+ if (isAutoDetectLanguage) {
+ val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
+ recognizer = SpeechRecognizer(speechConfig, autoDetectConfig, audioConfig)
+ } else {
+ recognizer = SpeechRecognizer(speechConfig, audioConfig)
+ }
+ } else {
+ // 使用默认麦克风输入
+ if (isAutoDetectLanguage) {
+ val autoDetectConfig = AutoDetectSourceLanguageConfig.fromLanguages(supportedLanguages.toList())
+ recognizer = SpeechRecognizer(speechConfig, autoDetectConfig)
+ } else {
+ recognizer = SpeechRecognizer(speechConfig)
+ }
+ }
+ return true
+ } else {
+ FileLogger.e(TAG, "语音配置未初始化")
+ return false
+ }
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "重置识别器失败: ${e.message}")
+ return false
+ }
+ }
+
+ // 开始一次性语音识别
+ fun recognizeOnce(callback: RecognizeCallback) {
+ if (speechConfig == null) {
+ callback.onError("语音服务未初始化")
+ return
+ }
+
+ // 重置 recognizer
+ if (!resetRecognizer()) {
+ callback.onError("重置识别器失败")
+ return
+ }
+
+ try {
+ // 启动音频处理
+ startCustomAudioProcessing()
+
+ // 执行识别
+ val result = recognizer?.recognizeOnceAsync()?.get()
+
+ // 停止音频处理
+ stopCustomAudioProcessing()
+
+ if (result != null && result.reason == ResultReason.RecognizedSpeech) {
+ val detectedLanguage = AutoDetectSourceLanguageResult.fromResult(result)?.language
+ callback.onResult(result.text, detectedLanguage ?: "")
+ } else {
+ callback.onError("未能识别语音")
+ }
+ } catch (e: Exception) {
+ // 停止音频处理
+ stopCustomAudioProcessing()
+ callback.onError("识别异常: ${e.message}")
+ }
+ }
+
+ // 开始连续语音识别
+ fun startContinuousRecognition(callback: ContinuousRecognizeCallback): Boolean {
+ if (speechConfig == null) {
+ callback.onError("语音服务未初始化")
+ return false
+ }
+
+ if (isContinuousRecognitionActive) {
+ FileLogger.w(TAG, "已在进行连续识别，忽略请求")
+ return true
+ }
+
+ // 重置 recognizer
+ if (!resetRecognizer()) {
+ callback.onError("重置识别器失败")
+ return false
+ }
+
+ try {
+ // 启动音频处理
+ startCustomAudioProcessing()
+
+ // 识别中事件
+ recognizer?.recognizing?.addEventListener(
+ EventHandler { _, event ->
+ val detectedLanguage = AutoDetectSourceLanguageResult.fromResult(event.result)?.language
+ FileLogger.d(TAG, "识别中: ${event.result.text}, 语言: $detectedLanguage")
+ callback.onRecognizing(event.result.text, detectedLanguage ?: "")
+ }
+ )
+
+ // 识别完成事件
+ recognizer?.recognized?.addEventListener(
+ EventHandler { _, event ->
+ if (event.result.reason == ResultReason.RecognizedSpeech) {
+ val detectedLanguage = AutoDetectSourceLanguageResult.fromResult(event.result)?.language
+ FileLogger.d(TAG, "识别完成: ${event.result.text}, 语言: $detectedLanguage")
+ callback.onResult(event.result.text, detectedLanguage ?: "")
+ }
+ }
+ )
+
+ // 会话开始事件
+ recognizer?.sessionStarted?.addEventListener(
+ EventHandler { _, _ ->
+ FileLogger.d(TAG, "识别会话已开始")
+ callback.onSessionStarted()
+ callback.onSuccess("开始识别") // 兼容旧接口
+ }
+ )
+
+ // 会话结束事件
+ recognizer?.sessionStopped?.addEventListener(
+ EventHandler { _, _ ->
+ FileLogger.d(TAG, "识别会话已结束")
+ isContinuousRecognitionActive = false
+ stopCustomAudioProcessing()
+ callback.onSessionStopped()
+ }
+ )
+
+ // 取消事件
+ recognizer?.canceled?.addEventListener(
+ EventHandler { _, event ->
+ val errorDetails = try {
+ event.errorDetails ?: "未知错误"
+ } catch (e: Exception) {
+ "未知错误"
+ }
+ val reason = event.reason.toString()
+ FileLogger.e(TAG, "识别取消: $errorDetails")
+ isContinuousRecognitionActive = false
+ stopCustomAudioProcessing()
+ callback.onCanceled(reason, errorDetails)
+ callback.onError("识别取消: $errorDetails") // 兼容旧接口
+ }
+ )
+
+ // 开始连续识别
+ recognizer?.startContinuousRecognitionAsync()
+ isContinuousRecognitionActive = true
+ FileLogger.d(TAG, "连续识别已启动")
+
+ return true
+ } catch (e: Exception) {
+ // 停止音频处理
+ stopCustomAudioProcessing()
+ FileLogger.e(TAG, "开始连续识别失败: ${e.message}")
+ e.printStackTrace()
+ callback.onError("开始连续识别失败: ${e.message}")
+ return false
+ }
+ }
+
+ // 停止连续语音识别
+ fun stopContinuousRecognition(callback: ContinuousRecognizeCallback): Boolean {
+ if (speechConfig == null) {
+ callback.onError("语音服务未初始化")
+ return false
+ }
+
+ if (!isContinuousRecognitionActive) {
+ FileLogger.d(TAG, "未进行连续识别，忽略停止请求")
+ return true
+ }
+
+ try {
+ FileLogger.d(TAG, "停止连续语音识别")
+
+ if (recognizer == null) {
+ if (isContinuousRecognitionActive) {
+ FileLogger.w(TAG, "识别器为空，但状态显示活跃")
+ }
+ isContinuousRecognitionActive = false
+ return true
+ }
+
+ // 停止连续识别
+ val future = recognizer?.stopContinuousRecognitionAsync()
+ future?.get()
+
+ // 停止音频处理
+ stopCustomAudioProcessing()
+
+ isContinuousRecognitionActive = false
+ FileLogger.d(TAG, "连续识别已停止")
+ callback.onSuccess("连续识别已停止")
+
+ return true
+ } catch (e: Exception) {
+ // 强制重置状态
+ isContinuousRecognitionActive = false
+ FileLogger.e(TAG, "停止连续识别失败: ${e.message}")
+ e.printStackTrace()
+ callback.onError("停止连续识别失败: ${e.message}")
+
+ // 停止音频处理
+ stopCustomAudioProcessing()
+
+ // 尝试强制关闭识别器
+ try {
+ recognizer?.close()
+ recognizer = null
+ } catch (ex: Exception) {
+ FileLogger.e(TAG, "关闭识别器失败: ${ex.message}")
+ }
+
+ return false
+ }
+ }
+
+ // 释放资源
+ fun dispose() {
+ try {
+ // 如果正在进行连续识别，先停止
+ if (isContinuousRecognitionActive) {
+ recognizer?.stopContinuousRecognitionAsync()?.get()
+ isContinuousRecognitionActive = false
+ }
+
+ // 停止音频处理
+ stopCustomAudioProcessing()
+
+ // 释放recognizer
+ recognizer?.close()
+ recognizer = null
+
+ // 释放speechConfig
+ speechConfig?.close()
+ speechConfig = null
+
+ FileLogger.d(TAG, "资源已释放")
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "释放资源失败: ${e.message}")
+
+ // 确保状态被重置
+ isContinuousRecognitionActive = false
+ customAudioProcessor = null
+ pushStream = null
+ audioConfig = null
+ recognizer = null
+ speechConfig = null
+ }
+ }
+
+ // 检查连续识别是否处于活跃状态
+ fun isContinuousRecognitionActive(): Boolean {
+ return isContinuousRecognitionActive
+ }
+
+ // 设置自定义音频处理
+ private fun setupCustomAudioProcessing() {
+ try {
+ // 创建音频推送流
+ pushStream = PushAudioInputStream.create()
+
+ // 创建音频配置
+ audioConfig = AudioConfig.fromStreamInput(pushStream)
+
+ // 创建自定义音频处理器
+ customAudioProcessor = CustomAudioProcessor(pushStream!!)
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置自定义音频处理失败: ${e.message}")
+ e.printStackTrace()
+ }
+ }
+
+ // 启动自定义音频处理
+ private fun startCustomAudioProcessing() {
+ if (useEchoCancellation && customAudioProcessor != null) {
+ try {
+ customAudioProcessor?.startProcessing()
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "启动音频处理器失败")
+ e.printStackTrace()
+ }
+ }
+ }
+
+ // 停止自定义音频处理
+ private fun stopCustomAudioProcessing() {
+ if (customAudioProcessor != null) {
+ try {
+ customAudioProcessor?.stopProcessing()
+ customAudioProcessor = null
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "停止音频处理器失败: ${e.message}")
+ e.printStackTrace()
+ }
+ }
+ }
+
+ // 修改认证取消事件处理代码
+ private fun setupCancelledEventHandler(callback: ContinuousRecognizeCallback) {
+ recognizer?.canceled?.addEventListener(
+ EventHandler { _, event ->
+ val errorDetails = try {
+ event.errorDetails ?: "未知错误"
+ } catch (e: Exception) {
+ "未知错误"
+ }
+ val reason = event.reason.toString()
+ FileLogger.e(TAG, "识别取消: $errorDetails")
+ isContinuousRecognitionActive = false
+ stopCustomAudioProcessing()
+ callback.onCanceled(reason, errorDetails)
+ callback.onError("识别取消: $errorDetails") // 兼容旧接口
+ }
+ )
+ }
+
+ // 自定义音频处理器
+ private inner class CustomAudioProcessor(private val pushStream: PushAudioInputStream) {
+ private val isProcessing = AtomicBoolean(false)
+ private var audioRecord: AudioRecord? = null
+ private var echoCanceler: AcousticEchoCanceler? = null
+ private var noiseSuppressor: NoiseSuppressor? = null
+ private var automaticGainControl: AutomaticGainControl? = null
+
+ // 音频配置
+ private val sampleRate = 16000 // 16kHz，适合语音识别
+ private val channelConfig = AudioFormat.CHANNEL_IN_MONO
+ private val audioFormat = AudioFormat.ENCODING_PCM_16BIT
+
+ // 计算最小 buffer 大小
+ private val bufferSize = AudioRecord.getMinBufferSize(
+ sampleRate, channelConfig, audioFormat
+ )
+
+ // 启动音频处理
+ fun startProcessing() {
+ if (isProcessing.get()) return
+
+ // 创建录音对象
+ try {
+ audioRecord = AudioRecord(
+ MediaRecorder.AudioSource.VOICE_RECOGNITION,
+ sampleRate,
+ channelConfig,
+ audioFormat,
+ bufferSize * 2 // 使用更大的缓冲区以确保不会丢失数据
+ )
+
+ // 创建音频处理效果
+ if (AcousticEchoCanceler.isAvailable()) {
+ echoCanceler = AcousticEchoCanceler.create(audioRecord!!.audioSessionId)
+ echoCanceler?.enabled = true
+ }
+
+ if (NoiseSuppressor.isAvailable()) {
+ noiseSuppressor = NoiseSuppressor.create(audioRecord!!.audioSessionId)
+ noiseSuppressor?.enabled = true
+ }
+
+ if (AutomaticGainControl.isAvailable()) {
+ automaticGainControl = AutomaticGainControl.create(audioRecord!!.audioSessionId)
+ automaticGainControl?.enabled = true
+ }
+
+ // 开始录音
+ audioRecord?.startRecording()
+
+ // 处理线程
+ Thread {
+ android.os.Process.setThreadPriority(Process.THREAD_PRIORITY_AUDIO)
+ processAudio()
+ }.start()
+
+ isProcessing.set(true)
+ FileLogger.d(TAG, "音频处理已启动")
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "创建音频处理器失败: ${e.message}")
+ releaseAudioResources()
+ throw e
+ }
+ }
+
+ // 停止音频处理
+ fun stopProcessing() {
+ if (!isProcessing.get()) return
+
+ isProcessing.set(false)
+ releaseAudioResources()
+ FileLogger.d(TAG, "音频处理已停止")
+ }
+
+ // 释放音频资源
+ private fun releaseAudioResources() {
+ try {
+ audioRecord?.stop()
+
+ echoCanceler?.release()
+ echoCanceler = null
+
+ noiseSuppressor?.release()
+ noiseSuppressor = null
+
+ automaticGainControl?.release()
+ automaticGainControl = null
+
+ audioRecord?.release()
+ audioRecord = null
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "释放音频资源失败: ${e.message}")
+ }
+ }
+
+ // 音频处理线程
+ private fun processAudio() {
+ // 设置线程优先级
+ try {
+ Process.setThreadPriority(Process.THREAD_PRIORITY_URGENT_AUDIO)
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置线程优先级失败")
+ }
+
+ val buffer = ByteArray(bufferSize)
+
+ while (isProcessing.get()) {
+ try {
+ val readSize = audioRecord?.read(buffer, 0, buffer.size) ?: -1
+
+ if (readSize > 0) {
+ // 修复: 只传入buffer，不传readSize
+ // 创建新的byte数组，只包含读取到的数据
+ val audioData = buffer.copyOfRange(0, readSize)
+ pushStream.write(audioData)
+ }
+
+ // 适当休眠，避免占用过多 CPU
+ Thread.sleep(5)
+ } catch (e: Exception) {
+ if (isProcessing.get()) {
+ FileLogger.e(TAG, "处理音频数据异常: ${e.message}")
+ }
+ break
+ }
+ }
+ }
+ }
+
+ // 一次性识别回调接口
+ interface RecognizeCallback {
+ fun onResult(text: String, detectedLanguage: String)
+ fun onError(error: String)
+ }
+
+ // 连续识别回调接口
+ interface ContinuousRecognizeCallback {
+ fun onResult(text: String, detectedLanguage: String)
+ fun onRecognizing(recognizing: String, detectedLanguage: String)
+ fun onSessionStarted()
+ fun onSessionStopped()
+ fun onCanceled(reason: String, errorDetails: String)
+ fun onError(error: String)
+
+ // 兼容旧版本的接口
+ fun onSuccess(message: String) {}
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureSpeechPlugin.kt b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureSpeechPlugin.kt
new file mode 100644
index 000000000..090a72efc
--- /dev/null
+++ b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureSpeechPlugin.kt
@@ -0,0 +1,297 @@
+package com.yunqiinnovation.azure_speech
+
+import android.content.Context
+import android.app.Activity
+import android.os.Handler
+import android.os.Looper
+import androidx.annotation.NonNull
+import com.yunqiinnovation.azure_speech.utils.FileLogger
+
+import io.flutter.embedding.engine.plugins.FlutterPlugin
+import io.flutter.plugin.common.MethodCall
+import io.flutter.plugin.common.MethodChannel
+import io.flutter.plugin.common.MethodChannel.MethodCallHandler
+import io.flutter.plugin.common.MethodChannel.Result
+import io.flutter.plugin.common.EventChannel
+
+/** AzureSpeechPlugin */
+class AzureSpeechPlugin: FlutterPlugin {
+ private val TAG = "AzureSpeechPlugin"
+ private lateinit var context: Context
+ private val mainHandler = Handler(Looper.getMainLooper())
+
+ // ASR相关
+ private lateinit var asrChannel : MethodChannel
+ private lateinit var asrEventChannel: EventChannel
+ private var asrEventSink: EventChannel.EventSink? = null
+ private lateinit var azureAsrHelper: AzureAsrHelper
+
+ // TTS相关
+ private lateinit var ttsChannel : MethodChannel
+ private lateinit var azureTtsHelper: AzureTtsHelper
+
+ // ASR 事件发送方法
+ private fun sendAsrEvent(event: Map) {
+ FileLogger.d(TAG, "发送ASR事件: $event")
+ if (asrEventSink == null) {
+ FileLogger.w(TAG, "无法发送ASR事件：事件通道未准备好")
+ return
+ }
+
+ mainHandler.post {
+ try {
+ asrEventSink?.success(event)
+ FileLogger.d(TAG, "ASR事件发送成功")
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "发送ASR事件失败: ${e.message}")
+ }
+ }
+ }
+
+ override fun onAttachedToEngine(@NonNull flutterPluginBinding: FlutterPlugin.FlutterPluginBinding) {
+ context = flutterPluginBinding.applicationContext
+
+ // 初始化ASR通道
+ asrChannel = MethodChannel(flutterPluginBinding.binaryMessenger, "azure_speech/asr")
+ asrChannel.setMethodCallHandler(AsrMethodHandler())
+
+ // 初始化TTS通道
+ ttsChannel = MethodChannel(flutterPluginBinding.binaryMessenger, "azure_speech/tts")
+ ttsChannel.setMethodCallHandler(TtsMethodHandler())
+
+ // 初始化ASR事件通道
+ asrEventChannel = EventChannel(flutterPluginBinding.binaryMessenger, "azure_speech/asr_events")
+ asrEventChannel.setStreamHandler(object : EventChannel.StreamHandler {
+ override fun onListen(arguments: Any?, events: EventChannel.EventSink?) {
+ asrEventSink = events
+ }
+
+ override fun onCancel(arguments: Any?) {
+ asrEventSink = null
+ }
+ })
+
+ // 初始化Azure语音服务
+ azureTtsHelper = AzureTtsHelper(context)
+ azureAsrHelper = AzureAsrHelper(context)
+ }
+
+ // ASR方法处理器
+ inner class AsrMethodHandler : MethodCallHandler {
+ override fun onMethodCall(@NonNull call: MethodCall, @NonNull result: Result) {
+ when (call.method) {
+ "initialize" -> {
+ val subscriptionKey = call.argument("subscriptionKey") ?: ""
+ val region = call.argument("region") ?: ""
+ val supportedLanguages = call.argument>("supportedLanguages") ?: listOf("zh-CN")
+
+ try {
+ val success = azureAsrHelper.initialize(subscriptionKey, region, supportedLanguages.toTypedArray())
+ result.success(success)
+ } catch (e: Exception) {
+ result.error("INITIALIZATION_ERROR", e.message, null)
+ }
+ }
+ "recognizeOnce" -> {
+ azureAsrHelper.recognizeOnce(object : AzureAsrHelper.RecognizeCallback {
+ override fun onResult(text: String, detectedLanguage: String) {
+ mainHandler.post {
+ result.success(mapOf(
+ "text" to text,
+ "detectedLanguage" to detectedLanguage
+ ))
+ }
+ }
+
+ override fun onError(error: String) {
+ mainHandler.post {
+ result.error("RECOGNITION_ERROR", error, null)
+ }
+ }
+ })
+ }
+ "startContinuousRecognition" -> {
+ // 确保事件通道已准备好
+ if (asrEventSink == null) {
+ result.error("EVENT_CHANNEL_NOT_READY", "事件通道未准备好，无法开始连续识别", null)
+ return
+ }
+
+ val success = azureAsrHelper.startContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
+ override fun onResult(text: String, detectedLanguage: String) {
+ sendAsrEvent(mapOf(
+ "type" to "result",
+ "text" to text,
+ "detectedLanguage" to detectedLanguage
+ ))
+ }
+
+ override fun onRecognizing(recognizing: String, detectedLanguage: String) {
+ sendAsrEvent(mapOf(
+ "type" to "recognizing",
+ "text" to recognizing,
+ "detectedLanguage" to detectedLanguage
+ ))
+ }
+
+ override fun onSessionStarted() {
+ sendAsrEvent(mapOf("type" to "sessionStarted"))
+ }
+
+ override fun onSessionStopped() {
+ sendAsrEvent(mapOf("type" to "sessionStopped"))
+ }
+
+ override fun onCanceled(reason: String, errorDetails: String) {
+ sendAsrEvent(mapOf(
+ "type" to "canceled",
+ "reason" to reason,
+ "errorDetails" to errorDetails
+ ))
+ }
+
+ override fun onError(error: String) {
+ sendAsrEvent(mapOf("type" to "error", "message" to error))
+ }
+
+ override fun onSuccess(message: String) {
+ sendAsrEvent(mapOf("type" to "success", "message" to message))
+ }
+ })
+ result.success(success)
+ }
+ "stopContinuousRecognition" -> {
+ try {
+ if (!azureAsrHelper.isContinuousRecognitionActive()) {
+ result.success(true)
+ return
+ }
+
+ val success = azureAsrHelper.stopContinuousRecognition(object : AzureAsrHelper.ContinuousRecognizeCallback {
+ override fun onResult(text: String, detectedLanguage: String) {}
+ override fun onRecognizing(recognizing: String, detectedLanguage: String) {}
+ override fun onSessionStarted() {}
+ override fun onSessionStopped() {}
+ override fun onCanceled(reason: String, errorDetails: String) {}
+ override fun onError(error: String) {
+ mainHandler.post {
+ result.error("STOP_ERROR", error, null)
+ }
+ }
+ override fun onSuccess(message: String) {}
+ })
+ result.success(success)
+ } catch (e: Exception) {
+ result.error("STOP_ERROR", e.message, null)
+ }
+ }
+ "isContinuousRecognitionActive" -> {
+ result.success(azureAsrHelper.isContinuousRecognitionActive())
+ }
+ "dispose" -> {
+ azureAsrHelper.dispose()
+ result.success(true)
+ }
+ else -> {
+ result.notImplemented()
+ }
+ }
+ }
+ }
+
+ // TTS方法处理器
+ inner class TtsMethodHandler : MethodCallHandler {
+ override fun onMethodCall(@NonNull call: MethodCall, @NonNull result: Result) {
+ when (call.method) {
+ "initialize" -> {
+ val subscriptionKey = call.argument("subscriptionKey") ?: ""
+ val region = call.argument("region") ?: ""
+ val language = call.argument("language") ?: "zh-CN"
+
+ val success = azureTtsHelper.initialize(subscriptionKey, region, language)
+ result.success(success)
+ }
+ "setVoice" -> {
+ val voiceName = call.argument("voiceName") ?: return result.error("INVALID_ARGUMENTS", "语音名称不能为空", null)
+ result.success(azureTtsHelper.setVoice(voiceName))
+ }
+ "setSpeechParams" -> {
+ val rate = call.argument("rate") ?: 0
+ val pitch = call.argument("pitch") ?: 0
+ val volume = call.argument("volume") ?: 100
+ result.success(azureTtsHelper.setSpeechParams(rate, pitch, volume))
+ }
+ "setAudioOutputType" -> {
+ val outputTypeStr = call.argument("outputType") ?: "speaker"
+ val outputType = when (outputTypeStr.lowercase()) {
+ "speaker" -> AzureTtsHelper.AudioOutputType.SPEAKER
+ "earpiece" -> AzureTtsHelper.AudioOutputType.EARPIECE
+ "auto" -> AzureTtsHelper.AudioOutputType.AUTO
+ else -> AzureTtsHelper.AudioOutputType.SPEAKER
+ }
+ result.success(azureTtsHelper.setAudioOutputType(outputType))
+ }
+ "speakText" -> {
+ val text = call.argument("text") ?: return result.error("INVALID_ARGUMENTS", "文本不能为空", null)
+
+ azureTtsHelper.speakText(text, object : AzureTtsHelper.TTSCallback {
+ override fun onSuccess(message: String) {
+ mainHandler.post {
+ result.success(true)
+ }
+ }
+
+ override fun onError(error: String) {
+ mainHandler.post {
+ result.error("SPEAK_ERROR", error, null)
+ }
+ }
+ })
+ }
+ "speakSsml" -> {
+ val ssml = call.argument("ssml") ?: return result.error("INVALID_ARGUMENTS", "SSML不能为空", null)
+
+ azureTtsHelper.speakSsml(ssml, object : AzureTtsHelper.TTSCallback {
+ override fun onSuccess(message: String) {
+ mainHandler.post {
+ result.success(true)
+ }
+ }
+
+ override fun onError(error: String) {
+ mainHandler.post {
+ result.error("SPEAK_ERROR", error, null)
+ }
+ }
+ })
+ }
+ "stopSpeaking" -> {
+ result.success(azureTtsHelper.stopSpeaking())
+ }
+ "isSpeaking" -> {
+ result.success(azureTtsHelper.isSpeaking())
+ }
+ "dispose" -> {
+ azureTtsHelper.dispose()
+ result.success(true)
+ }
+ else -> {
+ result.notImplemented()
+ }
+ }
+ }
+ }
+
+ override fun onDetachedFromEngine(@NonNull binding: FlutterPlugin.FlutterPluginBinding) {
+ asrChannel.setMethodCallHandler(null)
+ ttsChannel.setMethodCallHandler(null)
+ asrEventChannel.setStreamHandler(null)
+
+ try {
+ azureTtsHelper.dispose()
+ azureAsrHelper.dispose()
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "Dispose resources error: ${e.message}")
+ }
+ }
+}
\ No newline at end of file
diff --git a/android/app/src/main/kotlin/com/example/deep_voice/AzureTtsHelper.kt b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureTtsHelper.kt
similarity index 98%
rename from android/app/src/main/kotlin/com/example/deep_voice/AzureTtsHelper.kt
rename to local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureTtsHelper.kt
index 31971c2a4..0ae180fa9 100644
--- a/android/app/src/main/kotlin/com/example/deep_voice/AzureTtsHelper.kt
+++ b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/AzureTtsHelper.kt
@@ -1,11 +1,11 @@
-package com.yunqiinnovation.deepsound
+package com.yunqiinnovation.azure_speech
import android.content.Context
import android.media.AudioAttributes
import android.media.AudioDeviceInfo
import android.media.AudioManager
import android.os.Build
-import com.yunqiinnovation.deepsound.core.utils.FileLogger
+import com.yunqiinnovation.azure_speech.utils.FileLogger
import com.microsoft.cognitiveservices.speech.*
import com.microsoft.cognitiveservices.speech.audio.*
import java.util.concurrent.Future
@@ -50,16 +50,16 @@ class AzureTtsHelper(private val context: Context) {
* 初始化 TTS 引擎
*
* @param subscriptionKey Azure 语音服务订阅密钥
- * @param serviceRegion Azure 语音服务区域
+ * @param region Azure 语音服务区域
* @param language 可选，默认语言，默认为 "zh-CN"
*/
- fun initialize(subscriptionKey: String, serviceRegion: String, language: String = "zh-CN"): Boolean {
+ fun initialize(subscriptionKey: String, region: String, language: String = "zh-CN"): Boolean {
try {
audioManager = context.getSystemService(Context.AUDIO_SERVICE) as AudioManager
// 创建语音配置
- speechConfig = SpeechConfig.fromSubscription(subscriptionKey, serviceRegion)
+ speechConfig = SpeechConfig.fromSubscription(subscriptionKey, region)
// 设置语音合成输出格式为高质量音频
speechConfig?.setSpeechSynthesisOutputFormat(SpeechSynthesisOutputFormat.Riff24Khz16BitMonoPcm)
@@ -96,7 +96,7 @@ class AzureTtsHelper(private val context: Context) {
return true
} catch (e: Exception) {
- FileLogger.e(TAG, "TTS 引擎初始化失败: ${e.message}, ${serviceRegion}")
+ FileLogger.e(TAG, "TTS 引擎初始化失败: ${e.message}, ${region}")
e.printStackTrace()
return false
}
diff --git a/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/utils/FileLogger.kt b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/utils/FileLogger.kt
new file mode 100644
index 000000000..d210ec48f
--- /dev/null
+++ b/local_plugins/azure_speech/android/src/main/kotlin/com/yunqiinnovation/azure_speech/utils/FileLogger.kt
@@ -0,0 +1,45 @@
+package com.yunqiinnovation.azure_speech.utils
+
+import android.util.Log
+
+/**
+ * 简单的文件日志工具类
+ */
+object FileLogger {
+ private const val TAG_PREFIX = "AzureSpeech_"
+
+ /**
+ * 记录调试信息
+ */
+ fun d(tag: String, message: String) {
+ Log.d("$TAG_PREFIX$tag", message)
+ }
+
+ /**
+ * 记录信息
+ */
+ fun i(tag: String, message: String) {
+ Log.i("$TAG_PREFIX$tag", message)
+ }
+
+ /**
+ * 记录警告信息
+ */
+ fun w(tag: String, message: String) {
+ Log.w("$TAG_PREFIX$tag", message)
+ }
+
+ /**
+ * 记录错误信息
+ */
+ fun e(tag: String, message: String) {
+ Log.e("$TAG_PREFIX$tag", message)
+ }
+
+ /**
+ * 记录异常
+ */
+ fun e(tag: String, message: String, throwable: Throwable) {
+ Log.e("$TAG_PREFIX$tag", message, throwable)
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/azure_speech/ios/Classes/AzureAsrHelper.swift b/local_plugins/azure_speech/ios/Classes/AzureAsrHelper.swift
new file mode 100644
index 000000000..d3dd8d0b8
--- /dev/null
+++ b/local_plugins/azure_speech/ios/Classes/AzureAsrHelper.swift
@@ -0,0 +1,461 @@
+import Foundation
+import AVFoundation
+import MicrosoftCognitiveServicesSpeech
+
+/// Azure 语音识别辅助类
+class AzureAsrHelper: NSObject {
+ private var recognizer: SPXSpeechRecognizer?
+ private var speechConfig: SPXSpeechConfig?
+ private var audioConfig: SPXAudioConfig?
+ private var initialized = false
+ private var isContinuousRecognitionActive = false
+ private var currentLanguage = "zh-CN"
+ private var subscriptionKey = ""
+ private var serviceRegion = ""
+ private var isAutoDetectLanguage = false
+ private var supportedLanguages = ["zh-CN", "en-US"]
+
+ // 音频会话管理
+ private let audioSession = AVAudioSession.sharedInstance()
+
+ // 事件回调
+ private var eventHandler: (([String: Any]) -> Void)?
+
+ /// 设置事件处理器
+ ///
+ /// - Parameter handler: 事件处理回调
+ func setEventHandler(_ handler: @escaping ([String: Any]) -> Void) {
+ self.eventHandler = handler
+ }
+
+ /// 初始化语音识别服务
+ ///
+ /// - Parameters:
+ /// - speechSubscriptionKey: Azure 语音服务订阅密钥
+ /// - serviceRegion: Azure 语音服务区域
+ /// - supportedLanguages: 支持的语言列表，默认为 ["zh-CN", "en-US"]
+ /// - Returns: 是否初始化成功
+ func initialize(speechSubscriptionKey: String, serviceRegion: String, supportedLanguages: [String] = ["zh-CN", "en-US"]) -> Bool {
+ print("[AzureAsrHelper] 初始化 Azure 语音服务")
+
+ // 检查配置是否为空
+ if speechSubscriptionKey.isEmpty || serviceRegion.isEmpty {
+ print("[AzureAsrHelper] 错误: Azure 配置信息不完整")
+ return false
+ }
+
+ // 释放之前的资源
+ dispose()
+
+ // 保存配置
+ self.subscriptionKey = speechSubscriptionKey
+ self.serviceRegion = serviceRegion
+
+ // 设置语言
+ if supportedLanguages.isEmpty {
+ print("[AzureAsrHelper] 警告: 传入的支持语言列表为空，将使用默认语言")
+ } else {
+ self.supportedLanguages = supportedLanguages
+ }
+
+ // 根据支持的语言数量决定是否启用自动语言检测
+ self.isAutoDetectLanguage = supportedLanguages.count >= 2
+
+ // 如果只有一种语言，设置为当前语言
+ if !isAutoDetectLanguage && !supportedLanguages.isEmpty {
+ self.currentLanguage = supportedLanguages[0]
+ }
+
+ // 创建语音配置
+ do {
+ speechConfig = try SPXSpeechConfig(subscription: speechSubscriptionKey, region: serviceRegion)
+
+ // 设置语言配置
+ if isAutoDetectLanguage {
+ // 设置自动语言检测
+ try speechConfig?.setPropertyTo("Continuous", byId: SPXPropertyId.SpeechServiceConnection_LanguageIdMode)
+ } else {
+ // 设置指定的识别语言
+ speechConfig?.speechRecognitionLanguage = currentLanguage
+ }
+
+ // 创建音频配置 - 使用默认麦克风
+ audioConfig = SPXAudioConfig.default()
+
+ // 创建识别器
+ if isAutoDetectLanguage {
+ let autoDetectConfig = try SPXAutoDetectSourceLanguageConfiguration(sourceLanguages: supportedLanguages)
+ recognizer = try SPXSpeechRecognizer(speechConfiguration: speechConfig!, autoDetectSourceLanguageConfiguration: autoDetectConfig, audioConfiguration: audioConfig!)
+ } else {
+ recognizer = try SPXSpeechRecognizer(speechConfiguration: speechConfig!, audioConfiguration: audioConfig!)
+ }
+
+ // 配置音频会话
+ try configureAudioSession()
+
+ initialized = true
+ print("[AzureAsrHelper] Azure 语音服务初始化成功")
+ return true
+ } catch {
+ print("[AzureAsrHelper] 错误: 创建识别器失败: \(error.localizedDescription)")
+ return false
+ }
+ }
+
+ /// 配置音频会话
+ private func configureAudioSession() throws {
+ print("[AzureAsrHelper] 开始配置音频会话...")
+
+ do {
+ // 设置音频会话类别和模式
+ try audioSession.setCategory(.record, mode: .measurement, options: [.duckOthers, .allowBluetooth])
+ try audioSession.setActive(true, options: .notifyOthersOnDeactivation)
+ } catch {
+ print("[AzureAsrHelper] 警告: 通过AudioSessionManager配置音频会话失败")
+ throw error
+ }
+ }
+
+ /// 一次性语音识别
+ ///
+ /// - Parameter completion: 完成回调，返回是否成功、识别文本、识别语言和可能的错误信息
+ func recognizeOnce(completion: @escaping (Bool, String?, String?, String?) -> Void) {
+ if !initialized {
+ completion(false, nil, nil, "语音服务未初始化")
+ return
+ }
+
+ // 重置 recognizer
+ if !resetRecognizer() {
+ completion(false, nil, nil, "重置识别器失败")
+ return
+ }
+
+ do {
+ // 激活音频会话
+ try audioSession.setActive(true)
+
+ // 添加识别事件处理
+ recognizer?.addRecognizedEventHandler { [weak self] _, event in
+ guard let self = self else { return }
+
+ if event.result.reason == .recognizedSpeech {
+ let detectedLanguage = self.getDetectedLanguage(from: event.result)
+ print("[AzureAsrHelper] 识别结果: \(event.result.text ?? ""), 语言: \(detectedLanguage)")
+ completion(true, event.result.text, detectedLanguage, nil)
+ }
+ }
+
+ recognizer?.addRecognizingEventHandler { [weak self] _, event in
+ guard let self = self else { return }
+
+ if event.result.reason == .recognizingSpeech {
+ let detectedLanguage = self.getDetectedLanguage(from: event.result)
+ // print("[AzureAsrHelper] 识别中: \(event.result.text ?? ""), 语言: \(detectedLanguage)")
+ }
+ }
+
+ // 添加会话事件处理
+ recognizer?.addSessionStartedEventHandler { _, _ in
+ print("[AzureAsrHelper] 识别会话已开始")
+ }
+
+ recognizer?.addSessionStoppedEventHandler { _, _ in
+ print("[AzureAsrHelper] 识别会话已结束")
+ }
+
+ // 添加取消事件处理
+ recognizer?.addCanceledEventHandler { _, event in
+ if let cancellationDetails = try? SPXCancellationDetails(fromCanceledRecognitionResult: event.result) {
+ let errorDetails = cancellationDetails.errorDetails ?? "未知错误"
+ print("[AzureAsrHelper] 识别取消: \(errorDetails)")
+ completion(false, nil, nil, "识别取消: \(errorDetails)")
+ }
+ }
+
+ // 执行识别
+ let result = try recognizer?.recognizeOnceAsync().get()
+
+ if result?.reason != .recognizedSpeech {
+ completion(false, nil, nil, "未能识别语音")
+ }
+ } catch {
+ completion(false, nil, nil, "识别异常: \(error.localizedDescription)")
+ }
+ }
+
+ /// 重置识别器
+ ///
+ /// - Returns: 是否重置成功
+ private func resetRecognizer() -> Bool {
+ if !initialized {
+ print("[AzureAsrHelper] 错误: 语音服务未初始化")
+ return false
+ }
+
+ // 检查配置是否为空
+ if subscriptionKey.isEmpty || serviceRegion.isEmpty {
+ print("[AzureAsrHelper] 错误: Azure 配置信息不完整")
+ return false
+ }
+
+ do {
+ // 释放之前的 recognizer
+ recognizer = nil
+
+ // 创建音频配置 - 使用默认麦克风
+ audioConfig = SPXAudioConfig.default()
+
+ // 重新创建识别器
+ if isAutoDetectLanguage {
+ let autoDetectConfig = try SPXAutoDetectSourceLanguageConfiguration(sourceLanguages: supportedLanguages)
+ recognizer = try SPXSpeechRecognizer(speechConfiguration: speechConfig!, autoDetectSourceLanguageConfiguration: autoDetectConfig, audioConfiguration: audioConfig!)
+ } else {
+ recognizer = try SPXSpeechRecognizer(speechConfiguration: speechConfig!, audioConfiguration: audioConfig!)
+ }
+
+ return true
+ } catch {
+ print("[AzureAsrHelper] 错误: 重置识别器失败: \(error.localizedDescription)")
+ return false
+ }
+ }
+
+ /// 获取检测到的语言
+ ///
+ /// - Parameter result: 识别结果
+ /// - Returns: 检测到的语言代码
+ private func getDetectedLanguage(from result: SPXSpeechRecognitionResult) -> String {
+ if isAutoDetectLanguage {
+ do {
+ if let autoDetectResult = try SPXAutoDetectSourceLanguageResult(fromRecognitionResult: result) {
+ return autoDetectResult.language
+ }
+ return ""
+ } catch {
+ print("[AzureAsrHelper] 错误: 获取检测到的语言失败: \(error.localizedDescription)")
+ return ""
+ }
+ } else {
+ return currentLanguage
+ }
+ }
+
+ /// 开始连续语音识别
+ ///
+ /// - Returns: 是否成功启动连续识别
+ func startContinuousRecognition() -> Bool {
+ if !initialized {
+ print("[AzureAsrHelper] 错误: 语音服务未初始化")
+ return false
+ }
+
+ // 检查配置是否为空
+ if subscriptionKey.isEmpty || serviceRegion.isEmpty {
+ print("[AzureAsrHelper] 错误: Azure 配置信息不完整")
+ return false
+ }
+
+ // 如果已经在进行连续识别，直接返回
+ if isContinuousRecognitionActive {
+ print("[AzureAsrHelper] 已经在进行连续识别中，忽略请求")
+ return true
+ }
+
+ // 重置 recognizer
+ if !resetRecognizer() {
+ print("[AzureAsrHelper] 尝试重新创建识别器...")
+ return false
+ }
+
+ do {
+ // 激活音频会话
+ try audioSession.setActive(true)
+
+ // 添加识别事件处理
+ recognizer?.addRecognizedEventHandler { [weak self] _, event in
+ guard let self = self else { return }
+
+ if event.result.reason == .recognizedSpeech {
+ let detectedLanguage = self.getDetectedLanguage(from: event.result)
+ let eventData: [String: Any] = [
+ "eventType": "finalResult",
+ "text": event.result.text ?? "",
+ "language": detectedLanguage
+ ]
+ self.eventHandler?(eventData)
+ }
+ }
+
+ // 识别中事件
+ recognizer?.addRecognizingEventHandler { [weak self] _, event in
+ guard let self = self else { return }
+
+ if event.result.reason == .recognizingSpeech {
+ let detectedLanguage = self.getDetectedLanguage(from: event.result)
+ let eventData: [String: Any] = [
+ "eventType": "recognizing",
+ "text": event.result.text ?? "",
+ "language": detectedLanguage
+ ]
+ self.eventHandler?(eventData)
+ }
+ }
+
+ // 会话事件
+ recognizer?.addSessionStartedEventHandler { [weak self] _, _ in
+ guard let self = self else { return }
+
+ let eventData: [String: Any] = [
+ "eventType": "sessionStarted"
+ ]
+ self.eventHandler?(eventData)
+ }
+
+ recognizer?.addSessionStoppedEventHandler { [weak self] _, _ in
+ guard let self = self else { return }
+
+ self.isContinuousRecognitionActive = false
+ let eventData: [String: Any] = [
+ "eventType": "sessionStopped"
+ ]
+ self.eventHandler?(eventData)
+ }
+
+ // 取消事件
+ recognizer?.addCanceledEventHandler { [weak self] _, event in
+ guard let self = self else { return }
+
+ self.isContinuousRecognitionActive = false
+ var errorMessage = "未知错误"
+
+ if let cancellationDetails = try? SPXCancellationDetails(fromCanceledRecognitionResult: event.result) {
+ errorMessage = cancellationDetails.errorDetails ?? "未知错误"
+ }
+
+ let eventData: [String: Any] = [
+ "eventType": "error",
+ "error": "识别取消: \(errorMessage)"
+ ]
+ self.eventHandler?(eventData)
+ }
+
+ // 开始连续识别
+ try recognizer?.startContinuousRecognition()
+ isContinuousRecognitionActive = true
+ print("[AzureAsrHelper] 连续识别已启动")
+
+ return true
+ } catch {
+ print("[AzureAsrHelper] 错误: 开始连续识别失败: \(error.localizedDescription)")
+ return false
+ }
+ }
+
+ /// 停止连续语音识别
+ ///
+ /// - Returns: 是否成功停止连续识别
+ func stopContinuousRecognition() -> Bool {
+ if !initialized {
+ print("[AzureAsrHelper] 错误: 语音服务未初始化")
+ return false
+ }
+
+ if !isContinuousRecognitionActive {
+ print("[AzureAsrHelper] 未进行连续识别，忽略停止请求")
+ return true
+ }
+
+ do {
+ print("[AzureAsrHelper] 停止连续语音识别")
+
+ if recognizer == nil {
+ if isContinuousRecognitionActive {
+ print("[AzureAsrHelper] 警告: 识别器为空，但状态显示活跃")
+ }
+ isContinuousRecognitionActive = false
+
+ // 通知停止成功
+ let eventData: [String: Any] = [
+ "eventType": "success",
+ "message": "连续识别已停止"
+ ]
+ eventHandler?(eventData)
+ return true
+ }
+
+ // 停止连续识别
+ try recognizer?.stopContinuousRecognition()
+
+ // 延迟一点时间确保处理完成
+ DispatchQueue.main.asyncAfter(deadline: .now() + 0.5) { [weak self] in
+ guard let self = self else { return }
+
+ // 重置状态
+ self.isContinuousRecognitionActive = false
+
+ // 恢复音频会话
+ do {
+ try self.audioSession.setActive(false, options: .notifyOthersOnDeactivation)
+ } catch {
+ // 忽略错误
+ }
+
+ print("[AzureAsrHelper] 连续识别已停止")
+
+ // 通知停止成功
+ let eventData: [String: Any] = [
+ "eventType": "success",
+ "message": "连续识别已停止"
+ ]
+ self.eventHandler?(eventData)
+ }
+
+ return true
+ } catch {
+ // 强制重置状态
+ isContinuousRecognitionActive = false
+ print("[AzureAsrHelper] 警告: 停止连续识别失败: \(error.localizedDescription)")
+
+ // 通知停止失败，但仍然视为处理完成
+ let eventData: [String: Any] = [
+ "eventType": "success",
+ "message": "连续识别已停止（但有错误）"
+ ]
+ eventHandler?(eventData)
+
+ return false
+ }
+ }
+
+ /// 释放资源
+ func dispose() {
+ // 如果正在进行连续识别，先停止
+ if isContinuousRecognitionActive {
+ _ = stopContinuousRecognition()
+ }
+
+ // 恢复音频会话
+ do {
+ try audioSession.setActive(false, options: .notifyOthersOnDeactivation)
+ } catch {
+ // 忽略错误
+ }
+
+ // 释放资源
+ recognizer = nil
+ speechConfig = nil
+ audioConfig = nil
+
+ initialized = false
+ isContinuousRecognitionActive = false
+ print("[AzureAsrHelper] 资源已释放")
+ }
+
+ /// 检查连续识别是否处于活跃状态
+ ///
+ /// - Returns: 是否正在进行连续识别
+ func isContinuousRecognitionActive() -> Bool {
+ return isContinuousRecognitionActive
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/azure_speech/ios/Classes/AzureSpeechPlugin.swift b/local_plugins/azure_speech/ios/Classes/AzureSpeechPlugin.swift
new file mode 100644
index 000000000..dcb2d57b8
--- /dev/null
+++ b/local_plugins/azure_speech/ios/Classes/AzureSpeechPlugin.swift
@@ -0,0 +1,182 @@
+import Flutter
+import UIKit
+
+public class AzureSpeechPlugin: NSObject, FlutterPlugin {
+ private var ttsHelper: AzureTtsHelper?
+ private var asrHelper: AzureAsrHelper?
+ private var eventSink: FlutterEventSink?
+
+ public static func register(with registrar: FlutterPluginRegistrar) {
+ let channel = FlutterMethodChannel(name: "azure_speech", binaryMessenger: registrar.messenger())
+ let instance = AzureSpeechPlugin()
+ registrar.addMethodCallDelegate(instance, channel: channel)
+
+ // 初始化事件通道
+ let eventChannel = FlutterEventChannel(name: "azure_speech/asr_events", binaryMessenger: registrar.messenger())
+ eventChannel.setStreamHandler(AsrStreamHandler(instance: instance))
+ }
+
+ override init() {
+ super.init()
+ ttsHelper = AzureTtsHelper()
+ asrHelper = AzureAsrHelper()
+
+ // 设置ASR事件处理
+ asrHelper?.setEventHandler { [weak self] event in
+ self?.handleAsrEvent(event)
+ }
+ }
+
+ public func handle(_ call: FlutterMethodCall, result: @escaping FlutterResult) {
+ switch call.method {
+ // TTS相关方法
+ case "initializeTts":
+ guard let args = call.arguments as? [String: Any],
+ let subscriptionKey = args["subscriptionKey"] as? String,
+ let serviceRegion = args["serviceRegion"] as? String else {
+ result(false)
+ return
+ }
+
+ let language = args["language"] as? String ?? "zh-CN"
+ let success = ttsHelper?.initialize(speechSubscriptionKey: subscriptionKey, serviceRegion: serviceRegion, language: language) ?? false
+ result(success)
+
+ case "setTtsVoice":
+ guard let args = call.arguments as? [String: Any],
+ let voiceName = args["voiceName"] as? String else {
+ result(false)
+ return
+ }
+
+ let success = ttsHelper?.setVoice(voiceName: voiceName) ?? false
+ result(success)
+
+ case "setTtsSpeechParams":
+ guard let args = call.arguments as? [String: Any],
+ let rate = args["rate"] as? Int,
+ let pitch = args["pitch"] as? Int,
+ let volume = args["volume"] as? Int else {
+ result(false)
+ return
+ }
+
+ let success = ttsHelper?.setSpeechParams(rate: rate, pitch: pitch, volume: volume) ?? false
+ result(success)
+
+ case "speakText":
+ guard let args = call.arguments as? [String: Any],
+ let text = args["text"] as? String else {
+ result(false)
+ return
+ }
+
+ ttsHelper?.speakText(text: text) { success, _ in
+ result(success)
+ }
+
+ case "stopSpeaking":
+ let success = ttsHelper?.stopSpeaking() ?? false
+ result(success)
+
+ case "isSpeaking":
+ let speaking = ttsHelper?.isSpeaking() ?? false
+ result(speaking)
+
+ case "setTtsAudioOutputType":
+ guard let args = call.arguments as? [String: Any],
+ let outputType = args["outputType"] as? String else {
+ result(false)
+ return
+ }
+
+ var type: AzureTtsHelper.AudioOutputType = .auto
+ switch outputType.uppercased() {
+ case "SPEAKER":
+ type = .speaker
+ case "EARPIECE":
+ type = .earpiece
+ default:
+ type = .auto
+ }
+
+ let success = ttsHelper?.setAudioOutputType(outputType: type) ?? false
+ result(success)
+
+ // ASR相关方法
+ case "initializeAsr":
+ guard let args = call.arguments as? [String: Any],
+ let subscriptionKey = args["subscriptionKey"] as? String,
+ let serviceRegion = args["serviceRegion"] as? String,
+ let supportedLanguages = args["supportedLanguages"] as? [String] else {
+ result(false)
+ return
+ }
+
+ let success = asrHelper?.initialize(speechSubscriptionKey: subscriptionKey, serviceRegion: serviceRegion, supportedLanguages: supportedLanguages) ?? false
+ result(success)
+
+ case "recognizeOnce":
+ asrHelper?.recognizeOnce { success, text, language, error in
+ var resultMap: [String: Any] = ["success": success]
+ if success {
+ resultMap["text"] = text
+ resultMap["language"] = language
+ } else {
+ resultMap["error"] = error
+ }
+ result(resultMap)
+ }
+
+ case "startContinuousRecognition":
+ let success = asrHelper?.startContinuousRecognition() ?? false
+ result(success)
+
+ case "stopContinuousRecognition":
+ let success = asrHelper?.stopContinuousRecognition() ?? false
+ result(success)
+
+ case "isContinuousRecognitionActive":
+ let isActive = asrHelper?.isContinuousRecognitionActive() ?? false
+ result(isActive)
+
+ case "dispose":
+ ttsHelper?.dispose()
+ asrHelper?.dispose()
+ result(nil)
+
+ default:
+ result(FlutterMethodNotImplemented)
+ }
+ }
+
+ // 设置事件接收器
+ func setEventSink(_ sink: FlutterEventSink?) {
+ self.eventSink = sink
+ }
+
+ // 处理ASR事件
+ private func handleAsrEvent(_ event: [String: Any]) {
+ self.eventSink?(event)
+ }
+}
+
+// ASR事件流处理器
+class AsrStreamHandler: NSObject, FlutterStreamHandler {
+ private weak var plugin: AzureSpeechPlugin?
+
+ init(instance: AzureSpeechPlugin) {
+ self.plugin = instance
+ super.init()
+ }
+
+ func onListen(withArguments arguments: Any?, eventSink events: @escaping FlutterEventSink) -> FlutterError? {
+ plugin?.setEventSink(events)
+ return nil
+ }
+
+ func onCancel(withArguments arguments: Any?) -> FlutterError? {
+ plugin?.setEventSink(nil)
+ return nil
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/azure_speech/ios/Classes/AzureTtsHelper.swift b/local_plugins/azure_speech/ios/Classes/AzureTtsHelper.swift
new file mode 100644
index 000000000..4242a9037
--- /dev/null
+++ b/local_plugins/azure_speech/ios/Classes/AzureTtsHelper.swift
@@ -0,0 +1,330 @@
+import Foundation
+import AVFoundation
+import MicrosoftCognitiveServicesSpeech
+
+/// Azure 语音合成辅助类
+class AzureTtsHelper: NSObject {
+ private var synthesizer: SPXSpeechSynthesizer?
+ private var speechConfig: SPXSpeechConfig?
+ private var audioConfig: SPXAudioConfig?
+ private var initialized = false
+ private var speaking = false
+
+ // 音频输出类型
+ enum AudioOutputType {
+ case speaker // 扬声器
+ case earpiece // 听筒
+ case auto // 自动选择
+ }
+
+ // 当前设置
+ private var currentVoiceName = "zh-CN-XiaoxiaoNeural"
+ private var currentSpeechRate = 0
+ private var currentPitch = 0
+ private var currentVolume = 100
+ private var currentAudioOutputType: AudioOutputType = .auto
+
+ // 音频会话管理
+ private let audioSession = AVAudioSession.sharedInstance()
+
+ /// 初始化 TTS 引擎
+ ///
+ /// - Parameters:
+ /// - speechSubscriptionKey: Azure 语音服务订阅密钥
+ /// - serviceRegion: Azure 语音服务区域
+ /// - language: 可选，默认语言，默认为 "zh-CN"
+ /// - Returns: 是否初始化成功
+ func initialize(speechSubscriptionKey: String, serviceRegion: String, language: String = "zh-CN") -> Bool {
+ print("[AzureTtsHelper] 初始化 Azure 语音服务")
+
+ // 检查配置是否为空
+ if speechSubscriptionKey.isEmpty || serviceRegion.isEmpty {
+ print("[AzureTtsHelper] 错误: Azure 配置信息不完整")
+ return false
+ }
+
+ // 释放之前的资源
+ dispose()
+
+ do {
+ // 创建语音配置
+ speechConfig = try SPXSpeechConfig(subscription: speechSubscriptionKey, region: serviceRegion)
+
+ // 设置语音合成输出格式为高质量音频
+ speechConfig?.setSpeechSynthesisOutputFormat(.riff24Khz16BitMonoPcm)
+
+ // 设置默认语言
+ speechConfig?.setSpeechSynthesisLanguage(language)
+
+ // 设置默认语音
+ speechConfig?.setSpeechSynthesisVoiceName(currentVoiceName)
+
+ // 创建音频配置 - 使用默认扬声器
+ audioConfig = SPXAudioConfig.default()
+
+ // 创建语音合成器
+ synthesizer = try SPXSpeechSynthesizer(speechConfig: speechConfig!, audioConfig: audioConfig!)
+
+ initialized = true
+
+ // 设置默认音频输出类型为自动
+ setAudioOutputType(outputType: .auto)
+
+ print("[AzureTtsHelper] TTS 引擎初始化成功")
+ return true
+ } catch {
+ print("[AzureTtsHelper] TTS 引擎初始化失败: \(error.localizedDescription)")
+ return false
+ }
+ }
+
+ /// 设置音频输出设备类型
+ ///
+ /// - Parameter outputType: 音频输出设备类型
+ /// - Returns: 是否设置成功
+ func setAudioOutputType(outputType: AudioOutputType) -> Bool {
+ if !initialized {
+ print("[AzureTtsHelper] TTS 引擎尚未初始化")
+ return false
+ }
+
+ do {
+ currentAudioOutputType = outputType
+
+ switch outputType {
+ case .speaker:
+ // 使用扬声器
+ try audioSession.setCategory(.playback, mode: .default)
+ try audioSession.overrideOutputAudioPort(.speaker)
+ print("[AzureTtsHelper] 已设置音频输出设备为扬声器")
+
+ case .earpiece:
+ // 使用听筒
+ try audioSession.setCategory(.playback, mode: .voiceChat)
+ try audioSession.overrideOutputAudioPort(.none)
+ print("[AzureTtsHelper] 已设置音频输出设备为听筒")
+
+ case .auto:
+ // 检查是否有耳机连接
+ let outputs = audioSession.currentRoute.outputs
+ let hasHeadphones = outputs.contains { output in
+ return output.portType == .headphones || output.portType == .bluetoothA2DP || output.portType == .bluetoothHFP
+ }
+
+ if hasHeadphones {
+ // 有耳机，使用耳机
+ try audioSession.setCategory(.playback, mode: .default)
+ try audioSession.overrideOutputAudioPort(.none)
+ print("[AzureTtsHelper] 已设置音频输出设备为耳机")
+ } else {
+ // 无耳机，使用听筒
+ try audioSession.setCategory(.playback, mode: .voiceChat)
+ try audioSession.overrideOutputAudioPort(.none)
+ print("[AzureTtsHelper] 已设置音频输出设备为听筒")
+ }
+ }
+
+ try audioSession.setActive(true)
+ return true
+ } catch {
+ print("[AzureTtsHelper] 设置音频输出设备失败: \(error.localizedDescription)")
+ return false
+ }
+ }
+
+ /// 设置语音
+ ///
+ /// - Parameter voiceName: 语音名称，例如 "zh-CN-XiaoxiaoNeural"
+ /// - Returns: 是否设置成功
+ func setVoice(voiceName: String) -> Bool {
+ if !initialized {
+ print("[AzureTtsHelper] TTS 引擎尚未初始化")
+ return false
+ }
+
+ if voiceName == currentVoiceName {
+ print("[AzureTtsHelper] 已设置语音: \(voiceName)")
+ return true
+ }
+
+ do {
+ currentVoiceName = voiceName
+ speechConfig?.setSpeechSynthesisVoiceName(voiceName)
+
+ // 重新创建合成器
+ synthesizer = try SPXSpeechSynthesizer(speechConfig: speechConfig!, audioConfig: audioConfig!)
+
+ print("[AzureTtsHelper] 已设置语音: \(voiceName)")
+ return true
+ } catch {
+ print("[AzureTtsHelper] 设置语音失败: \(error.localizedDescription)")
+ return false
+ }
+ }
+
+ /// 设置语音合成参数
+ ///
+ /// - Parameters:
+ /// - rate: 语速，范围 -100 到 100，默认为 0
+ /// - pitch: 音调，范围 -100 到 100，默认为 0
+ /// - volume: 音量，范围 0 到 100，默认为 100
+ /// - Returns: 是否设置成功
+ func setSpeechParams(rate: Int = 0, pitch: Int = 0, volume: Int = 100) -> Bool {
+ if !initialized {
+ print("[AzureTtsHelper] TTS 引擎尚未初始化")
+ return false
+ }
+
+ currentSpeechRate = rate
+ currentPitch = pitch
+ currentVolume = volume
+
+ print("[AzureTtsHelper] 已设置语音参数: 语速=\(rate), 音调=\(pitch), 音量=\(volume)")
+ return true
+ }
+
+ /// 合成文本为语音并播放
+ ///
+ /// - Parameters:
+ /// - text: 要合成的文本
+ /// - completion: 完成回调，返回是否成功和可能的错误信息
+ func speakText(text: String, completion: @escaping (Bool, String?) -> Void) {
+ if !initialized {
+ print("[AzureTtsHelper] TTS 引擎尚未初始化")
+ completion(false, "TTS 引擎尚未初始化")
+ return
+ }
+
+ do {
+ print("[AzureTtsHelper] 开始合成文本: \(text)")
+
+ // 生成 SSML
+ let ssml = generateSsml(text: text)
+
+ // 使用 SSML 合成语音
+ speakSsml(ssml: ssml, completion: completion)
+ } catch {
+ print("[AzureTtsHelper] 语音合成异常: \(error.localizedDescription)")
+ completion(false, "语音合成异常: \(error.localizedDescription)")
+ }
+ }
+
+ /// 生成 SSML 文本
+ ///
+ /// - Parameter text: 要转换的文本
+ /// - Returns: SSML 格式的文本
+ private func generateSsml(text: String) -> String {
+ // 计算 SSML 参数
+ let rateParam = currentSpeechRate == 0 ? "0%" : (currentSpeechRate < 0 ? "\(Int(Double(currentSpeechRate) * 0.9))%" : "\(currentSpeechRate)%")
+ let pitchParam = currentPitch == 0 ? "0%" : "\(Int(Double(currentPitch) * 0.5))%"
+ let volumeParam = "\(min(max(currentVolume, 0), 100))%"
+
+ return """
+
+
+
+ \(text)
+
+
+
+ """
+ }
+
+ /// 合成 SSML 为语音并播放
+ ///
+ /// - Parameters:
+ /// - ssml: SSML 格式的文本
+ /// - completion: 完成回调，返回是否成功和可能的错误信息
+ private func speakSsml(ssml: String, completion: @escaping (Bool, String?) -> Void) {
+ if !initialized {
+ print("[AzureTtsHelper] TTS 引擎尚未初始化")
+ completion(false, "TTS 引擎尚未初始化")
+ return
+ }
+
+ do {
+ print("[AzureTtsHelper] 开始合成 SSML")
+
+ // 标记为正在播放
+ speaking = true
+
+ // 激活音频会话
+ try audioSession.setActive(true)
+
+ // 异步合成语音
+ let result = try synthesizer!.speakSsml(ssml)
+
+ switch result.reason {
+ case .synthesizingAudioCompleted:
+ print("[AzureTtsHelper] 语音合成完成")
+ speaking = false
+ completion(true, "语音合成完成")
+ case .canceled:
+ if let cancelDetails = try? SPXSpeechSynthesisCancellationDetails(fromResult: result) {
+ print("[AzureTtsHelper] 语音合成取消: \(cancelDetails.errorDetails ?? "未知错误")")
+ speaking = false
+ completion(false, "语音合成取消: \(cancelDetails.errorDetails ?? "未知错误")")
+ } else {
+ print("[AzureTtsHelper] 语音合成取消")
+ speaking = false
+ completion(false, "语音合成取消")
+ }
+ default:
+ print("[AzureTtsHelper] 语音合成失败: \(result.reason)")
+ speaking = false
+ completion(false, "语音合成失败: \(result.reason)")
+ }
+ } catch {
+ print("[AzureTtsHelper] 语音合成异常: \(error.localizedDescription)")
+ speaking = false
+ completion(false, "语音合成异常: \(error.localizedDescription)")
+ }
+ }
+
+ /// 停止当前语音合成
+ ///
+ /// - Returns: 是否停止成功
+ func stopSpeaking() -> Bool {
+ if !initialized {
+ print("[AzureTtsHelper] TTS 引擎尚未初始化")
+ return false
+ }
+
+ do {
+ try synthesizer?.stopSpeaking()
+ speaking = false
+ print("[AzureTtsHelper] 已停止语音合成")
+ return true
+ } catch {
+ print("[AzureTtsHelper] 停止语音合成失败: \(error.localizedDescription)")
+ return false
+ }
+ }
+
+ /// 释放资源
+ func dispose() {
+ do {
+ stopSpeaking()
+
+ // 恢复音频会话
+ try audioSession.setActive(false, options: .notifyOthersOnDeactivation)
+
+ synthesizer = nil
+ speechConfig = nil
+ audioConfig = nil
+
+ initialized = false
+ speaking = false
+ print("[AzureTtsHelper] TTS 引擎已释放")
+ } catch {
+ print("[AzureTtsHelper] 释放 TTS 引擎失败: \(error.localizedDescription)")
+ }
+ }
+
+ /// 检查当前是否正在播放语音
+ ///
+ /// - Returns: 是否正在播放语音
+ func isSpeaking() -> Bool {
+ return speaking
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/azure_speech/pubspec.yaml b/local_plugins/azure_speech/pubspec.yaml
new file mode 100644
index 000000000..5935bbce2
--- /dev/null
+++ b/local_plugins/azure_speech/pubspec.yaml
@@ -0,0 +1,32 @@
+name: azure_speech
+description: Azure语音服务插件，包含TTS和ASR服务
+version: 0.0.1
+homepage:
+
+environment:
+ sdk: ">=2.17.0 <3.0.0"
+ flutter: ">=2.5.0"
+
+dependencies:
+ flutter:
+ sdk: flutter
+
+dev_dependencies:
+ flutter_test:
+ sdk: flutter
+ flutter_lints: ^2.0.0
+
+# For information on the generic Dart part of this file, see the
+# following page: https://dart.dev/tools/pub/pubspec
+
+# The following section is specific to Flutter packages.
+flutter:
+ # This section identifies this Flutter project as a plugin project.
+ plugin:
+ platforms:
+ android:
+ package: com.yunqiinnovation.azure_speech
+ pluginClass: AzureSpeechPlugin
+ ios:
+ pluginClass: AzureSpeechPlugin
+
diff --git a/local_plugins/open_ai_service/README.md b/local_plugins/open_ai_service/README.md
new file mode 100644
index 000000000..351eee84f
--- /dev/null
+++ b/local_plugins/open_ai_service/README.md
@@ -0,0 +1,253 @@
+# OpenAI Service Plugin
+
+一个用于Flutter应用的OpenAI服务插件，支持Android和iOS平台。
+
+## 功能
+
+- 支持文本生成（completions）
+- 支持流式输出（streaming）
+- 支持函数调用（function calling）
+- 支持自定义API基础URL
+- 支持自定义模型选择
+
+## 安装
+
+在你的`pubspec.yaml`文件中添加以下依赖：
+
+```yaml
+dependencies:
+ open_ai_service:
+ path: 本地路径/open_ai_service
+```
+
+## 使用方法
+
+### 初始化服务
+
+```dart
+import 'package:open_ai_service/open_ai_service.dart';
+
+final openAIService = OpenAIService();
+
+// 初始化服务
+await openAIService.initialize(
+ apiKey: 'your_openai_api_key',
+ baseUrl: 'https://api.openai.com/v1/chat/completions', // 可选
+ model: 'gpt-4-turbo', // 可选
+);
+```
+
+### 发送非流式请求
+
+```dart
+// 创建消息
+final userMessage = await openAIService.createUserMessage('你好，请介绍一下自己');
+
+// 发送请求
+final response = await openAIService.sendMessage(
+ messages: [userMessage],
+ systemPrompt: '你是一个有用的AI助手',
+);
+
+print('AI回复: $response');
+```
+
+### 发送流式请求（回调方式）
+
+```dart
+// 创建消息
+final userMessage = await openAIService.createUserMessage('写一个短故事');
+
+// 发送流式请求
+await openAIService.sendMessageStream(
+ messages: [userMessage],
+ systemPrompt: '你是一个善于讲故事的AI助手',
+);
+
+// 处理事件
+final subscription = openAIService.processEvents(
+ onToken: (token) {
+ // 处理每个返回的token
+ print(token);
+ },
+ onComplete: () {
+ // 处理完成事件
+ print('生成完成');
+ },
+ onError: (error) {
+ // 处理错误
+ print('错误: $error');
+ },
+ onFunctionCall: (functionCall) {
+ // 处理函数调用
+ print('函数调用: ${functionCall['name']}');
+ },
+);
+
+// 在不需要时取消订阅
+subscription.cancel();
+```
+
+### 发送流式请求（Stream方式）
+
+```dart
+// 创建消息
+final userMessage = await openAIService.createUserMessage('写一个短故事');
+
+// 获取字符串流
+final stream = openAIService.streamMessage(
+ messages: [userMessage],
+ systemPrompt: '你是一个善于讲故事的AI助手',
+);
+
+// 使用流
+final StringBuilder responseBuilder = StringBuilder();
+
+stream.listen(
+ (token) {
+ // 处理每个token
+ responseBuilder.write(token);
+ print(token); // 实时输出
+ },
+ onDone: () {
+ // 流结束
+ print('完整回复: ${responseBuilder.toString()}');
+ },
+ onError: (error) {
+ // 错误处理
+ print('错误: $error');
+ }
+);
+```
+
+### 注册函数
+
+```dart
+// 注册一个函数
+await openAIService.registerFunction(
+ name: 'get_weather',
+ description: '获取指定城市的天气信息',
+ parameters: {
+ 'type': 'object',
+ 'properties': {
+ 'city': {
+ 'type': 'string',
+ 'description': '城市名称',
+ },
+ 'date': {
+ 'type': 'string',
+ 'description': '日期，格式为YYYY-MM-DD',
+ },
+ },
+ 'required': ['city'],
+ },
+);
+```
+
+### 处理函数调用
+
+```dart
+// 创建消息
+final userMessage = await openAIService.createUserMessage('明天北京的天气如何？');
+
+// 发送流式请求
+await openAIService.sendMessageStream(
+ messages: [userMessage],
+ systemPrompt: '你是一个有用的AI助手',
+);
+
+// 处理事件
+openAIService.processEvents(
+ onToken: (token) {
+ print(token);
+ },
+ onComplete: () {
+ print('生成完成');
+ },
+ onError: (error) {
+ print('错误: $error');
+ },
+ onFunctionCall: (functionCall) {
+ // 处理函数调用
+ final name = functionCall['name'];
+ final arguments = functionCall['arguments'];
+
+ print('收到函数调用: $name, 参数: $arguments');
+
+ // 假设处理了函数调用并获得结果
+ final result = '{"temperature": 25, "condition": "sunny"}';
+
+ // 发送函数调用结果
+ openAIService.sendFunctionCallResult(
+ messages: [userMessage],
+ systemPrompt: '你是一个有用的AI助手',
+ functionCall: functionCall,
+ functionResult: result,
+ );
+ },
+);
+```
+
+### 使用Stream API处理函数调用
+
+```dart
+// 创建消息和响应处理器
+final userMessage = await openAIService.createUserMessage('明天北京的天气如何？');
+final responseBuilder = StringBuilder();
+
+// 处理事件流以捕获函数调用
+final subscription = openAIService.processEvents(
+ onFunctionCall: (functionCall) async {
+ // 取消当前事件监听
+ subscription.cancel();
+
+ // 处理函数调用
+ final name = functionCall['name'];
+ final arguments = functionCall['arguments'];
+
+ print('收到函数调用: $name, 参数: $arguments');
+
+ // 假设处理了函数调用并获得结果
+ final result = '{"temperature": 25, "condition": "sunny"}';
+
+ // 使用Stream API发送函数调用结果
+ final resultStream = openAIService.streamFunctionResult(
+ messages: [userMessage],
+ systemPrompt: '你是一个有用的AI助手',
+ functionCall: functionCall,
+ functionResult: result,
+ );
+
+ // 处理结果流
+ resultStream.listen(
+ (token) {
+ responseBuilder.write(token);
+ print(token); // 实时输出
+ },
+ onDone: () {
+ print('完整回复: ${responseBuilder.toString()}');
+ },
+ onError: (error) {
+ print('错误: $error');
+ }
+ );
+ }
+);
+
+// 启动请求
+await openAIService.sendMessageStream(
+ messages: [userMessage],
+ systemPrompt: '你是一个有用的AI助手',
+);
+```
+
+## 注意事项
+
+1. 确保在使用前已正确初始化服务
+2. 对于流式请求，确保在不需要时取消订阅
+3. 处理函数调用时，确保提供有效的结果格式
+4. 网络请求可能会失败，请确保加入适当的错误处理
+
+## 许可证
+
+[MIT License](LICENSE)
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/android/build.gradle.kts b/local_plugins/open_ai_service/android/build.gradle.kts
new file mode 100644
index 000000000..4f7b2f643
--- /dev/null
+++ b/local_plugins/open_ai_service/android/build.gradle.kts
@@ -0,0 +1,37 @@
+plugins {
+ // Android Library 插件
+ id("com.android.library")
+ // Kotlin Android 插件
+ id("org.jetbrains.kotlin.android")
+}
+
+android {
+ // 命名空间，对应你插件的包名（需与代码内包名保持一致）
+ namespace = "com.yunqiinnovation.open_ai_service"
+
+ // 目标 SDK 版本
+ compileSdk = 33
+
+ defaultConfig {
+ // 最低 SDK 版本
+ minSdk = 21
+ targetSdk = 33
+ }
+
+ // Java 语言级别兼容配置
+ compileOptions {
+ sourceCompatibility = JavaVersion.VERSION_11
+ targetCompatibility = JavaVersion.VERSION_11
+ }
+
+ // Kotlin 语言级别
+ kotlinOptions {
+ jvmTarget = "11"
+ }
+}
+
+dependencies {
+
+ implementation("com.squareup.okhttp3:okhttp:4.10.0")
+
+}
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/android/settings.gradle.kts b/local_plugins/open_ai_service/android/settings.gradle.kts
new file mode 100644
index 000000000..1c33f4971
--- /dev/null
+++ b/local_plugins/open_ai_service/android/settings.gradle.kts
@@ -0,0 +1 @@
+rootProject.name = "open_ai_service"
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/android/src/main/AndroidManifest.xml b/local_plugins/open_ai_service/android/src/main/AndroidManifest.xml
new file mode 100644
index 000000000..e38cdeb21
--- /dev/null
+++ b/local_plugins/open_ai_service/android/src/main/AndroidManifest.xml
@@ -0,0 +1,11 @@
+
+
+
+
+
+
+
+
+
+
diff --git a/local_plugins/open_ai_service/android/src/main/kotlin/com/yunqiinnovation/open_ai_service/OpenAIService.kt b/local_plugins/open_ai_service/android/src/main/kotlin/com/yunqiinnovation/open_ai_service/OpenAIService.kt
new file mode 100644
index 000000000..907be93a2
--- /dev/null
+++ b/local_plugins/open_ai_service/android/src/main/kotlin/com/yunqiinnovation/open_ai_service/OpenAIService.kt
@@ -0,0 +1,469 @@
+package com.yunqiinnovation.open_ai_service
+
+import android.util.Log
+import okhttp3.*
+import okhttp3.MediaType.Companion.toMediaTypeOrNull
+import okhttp3.RequestBody.Companion.toRequestBody
+import org.json.JSONArray
+import org.json.JSONObject
+import java.io.IOException
+import java.util.concurrent.TimeUnit
+
+/**
+ * OpenAI服务的原生实现
+ */
+class OpenAIService() {
+ private val TAG = "OpenAIService"
+ private var baseUrl = ""
+ private val client = OkHttpClient.Builder()
+ .connectTimeout(30, TimeUnit.SECONDS)
+ .readTimeout(30, TimeUnit.SECONDS)
+ .writeTimeout(30, TimeUnit.SECONDS)
+ .build()
+
+ private var apiKey: String = ""
+ private var isInitialized = false
+ private var model: String = "" // 默认模型
+
+ // 用于存储注册的函数
+ private val registeredFunctions = mutableListOf()
+
+ /**
+ * 构建curl命令用于测试
+ */
+ private fun buildCurlCommand(request: Request, body: String): String {
+ val command = StringBuilder("curl -v -X ${request.method}")
+
+ // 添加请求头
+ request.headers.forEach { header ->
+ // 敏感信息处理：不显示真实的API Key
+ if (header.first == "Authorization") {
+ command.append(" -H '${header.first}: Bearer $apiKey'")
+ } else {
+ command.append(" -H '${header.first}: ${header.second}'")
+ }
+ }
+
+ // 添加请求体
+ if (request.method == "POST" || request.method == "PUT") {
+ // 转义JSON中的单引号，确保curl命令正确
+ val escapedBody = body.replace("'", "\\'")
+ command.append(" -d '${escapedBody}'")
+ }
+
+ // 添加URL
+ command.append(" '${request.url}'")
+
+ return command.toString()
+ }
+
+ /**
+ * 创建用户消息
+ */
+ fun createUserMessage(content: String): JSONObject {
+ return JSONObject().apply {
+ put("role", "user")
+ put("content", content)
+ }
+ }
+
+ /**
+ * 创建助手消息
+ */
+ fun createAssistantMessage(content: String): JSONObject {
+ return JSONObject().apply {
+ put("role", "assistant")
+ put("content", content)
+ }
+ }
+
+ /**
+ * 初始化OpenAI服务
+ */
+ fun initialize(apiKey: String, baseUrl: String, model: String): Boolean {
+ this.apiKey = apiKey
+ if (baseUrl.isNotEmpty()) {
+ this.baseUrl = baseUrl
+ }
+ if (model.isNotEmpty()) {
+ this.model = model
+ }
+ isInitialized = apiKey.isNotEmpty()
+ return isInitialized
+ }
+
+ /**
+ * 注册函数
+ */
+ fun registerFunction(name: String, description: String, parameters: JSONObject): Boolean {
+ try {
+ val function = JSONObject().apply {
+ put("name", name)
+ put("description", description)
+ put("parameters", parameters)
+ }
+
+ // 检查是否已存在相同名称的函数
+ val existingIndex = registeredFunctions.indexOfFirst {
+ it.getString("name") == name
+ }
+
+ if (existingIndex >= 0) {
+ // 如果已存在，则替换
+ registeredFunctions[existingIndex] = function
+ } else {
+ // 如果不存在，则添加
+ registeredFunctions.add(function)
+ }
+
+ return true
+ } catch (e: Exception) {
+ return false
+ }
+ }
+
+ /**
+ * 发送消息（非流式输出）
+ */
+ @Throws(OpenAIException::class)
+ fun sendMessage(messages: JSONArray): String {
+ if (!isInitialized || apiKey.isEmpty()) {
+ throw OpenAIException("OpenAI服务未初始化")
+ }
+
+ val requestBody = JSONObject().apply {
+ put("model", model)
+ put("messages", messages)
+ put("temperature", 0.7)
+ put("max_tokens", 2000)
+ put("stream", false)
+
+ // 如果有注册的函数，则添加到请求中
+ if (registeredFunctions.isNotEmpty()) {
+ val tools = JSONArray()
+ for (function in registeredFunctions) {
+ val tool = JSONObject().apply {
+ put("type", "function")
+ put("function", function)
+ }
+ tools.put(tool)
+ }
+ put("tools", tools)
+ }
+ }
+
+ val mediaType = "application/json".toMediaTypeOrNull()
+ val request = Request.Builder()
+ .url(baseUrl)
+ .addHeader("Content-Type", "application/json")
+ .addHeader("Authorization", "Bearer $apiKey")
+ .post(requestBody.toString().toRequestBody(mediaType))
+ .build()
+
+ try {
+ // 输出用于测试的curl命令
+ // val curlCommand = buildCurlCommand(request, requestBody.toString())
+ // Log.d(TAG, "curl command: \n$curlCommand")
+
+ client.newCall(request).execute().use { response ->
+ if (!response.isSuccessful) {
+ throw OpenAIException("API调用失败: ${response.code}")
+ }
+
+ val responseBody = response.body?.string() ?: throw OpenAIException("Empty response")
+ val jsonResponse = JSONObject(responseBody)
+
+ // 检查是否有函数调用
+ if (jsonResponse.has("choices") &&
+ jsonResponse.getJSONArray("choices").length() > 0) {
+
+ val choice = jsonResponse.getJSONArray("choices").getJSONObject(0)
+
+ // 检查是否是函数调用
+ if (choice.has("message")) {
+ val message = choice.getJSONObject("message")
+
+ // 检查是否有工具调用
+ if (message.has("tool_calls")) {
+ val toolCalls = message.getJSONArray("tool_calls")
+ if (toolCalls.length() > 0) {
+ val toolCall = toolCalls.getJSONObject(0)
+ if (toolCall.has("function")) {
+ val function = toolCall.getJSONObject("function")
+ val functionCall = JSONObject().apply {
+ put("name", function.getString("name"))
+ put("arguments", function.getString("arguments"))
+ put("id", toolCall.getString("id"))
+ }
+ return functionCall.toString()
+ }
+ }
+ }
+
+ // 如果没有工具调用，返回消息内容
+ if (message.has("content")) {
+ return message.getString("content")
+ }
+ }
+ }
+
+ throw OpenAIException("Invalid response format")
+ }
+ } catch (e: Exception) {
+ if (e is OpenAIException) throw e
+ throw OpenAIException("Failed to communicate with AI service: ${e.message}")
+ }
+ }
+
+ /**
+ * 发送消息（流式输出）
+ */
+ fun sendMessageStream(messages: JSONArray, callback: StreamCallback) {
+ if (!isInitialized || apiKey.isEmpty()) {
+ callback.onError(OpenAIException("OpenAI服务未初始化"))
+ return
+ }
+
+ val requestBody = JSONObject().apply {
+ put("model", model)
+ put("messages", messages)
+ put("temperature", 0.7)
+ put("max_tokens", 2000)
+ put("stream", true)
+
+ // 如果有注册的函数，则添加到请求中
+ if (registeredFunctions.isNotEmpty()) {
+ val tools = JSONArray()
+ for (function in registeredFunctions) {
+ val tool = JSONObject().apply {
+ put("type", "function")
+ put("function", function)
+ }
+ tools.put(tool)
+ }
+ put("tools", tools)
+ }
+ }
+
+ val mediaType = "application/json".toMediaTypeOrNull()
+ val request = Request.Builder()
+ .url(baseUrl)
+ .addHeader("Content-Type", "application/json")
+ .addHeader("Authorization", "Bearer $apiKey")
+ .addHeader("Accept", "text/event-stream")
+ .post(requestBody.toString().toRequestBody(mediaType))
+ .build()
+
+ // 输出用于测试的curl命令
+ // val curlCommand = buildCurlCommand(request, requestBody.toString())
+ // Log.d(TAG, "curl command: $curlCommand")
+
+ client.newCall(request).enqueue(object : Callback {
+ override fun onFailure(call: Call, e: IOException) {
+ callback.onError(OpenAIException(e.message ?: "请求失败"))
+ }
+
+ override fun onResponse(call: Call, response: Response) {
+
+ if (!response.isSuccessful) {
+ callback.onError(OpenAIException("API调用失败: ${response.code}"))
+ return
+ }
+
+ val responseBody = response.body ?: return
+ val source = responseBody.source()
+
+ try {
+ // 预取数据到缓冲区
+ source.request(Long.MAX_VALUE)
+ val bufferedSource = source.buffer
+
+ // 用于存储函数调用的各个部分
+ val finalToolCalls = mutableMapOf()
+
+ while (!bufferedSource.exhausted()) {
+ val line = bufferedSource.readUtf8Line()?.trim() ?: continue
+ if (line.isEmpty()) continue
+ if (line.startsWith("data:")) {
+ val data = line.substring(5).trim()
+
+ // 处理[DONE]消息
+ if (data == "[DONE]" || data == "[\"DONE\"]") {
+ processToolCalls(finalToolCalls, callback)
+ callback.onComplete()
+ break
+ }
+
+ try {
+ val jsonData = JSONObject(data)
+
+ // 处理消息内容
+ if (jsonData.has("choices")) {
+ val choices = jsonData.getJSONArray("choices")
+ if (choices.length() > 0) {
+ val choice = choices.getJSONObject(0)
+
+ if (choice.has("delta")) {
+ val delta = choice.getJSONObject("delta")
+
+ // 处理普通文本内容
+ if (delta.has("content")) {
+ val content = delta.getString("content")
+ callback.onToken(content)
+ }
+
+ // 处理工具调用（函数调用）
+ if (delta.has("tool_calls")) {
+ val toolCalls = delta.getJSONArray("tool_calls")
+ for (i in 0 until toolCalls.length()) {
+ val toolCall = toolCalls.getJSONObject(i)
+ val index = toolCall.getInt("index")
+
+ // 创建或获取现有的工具调用信息
+ val toolCallInfo = finalToolCalls.getOrPut(index) { ToolCallInfo() }
+
+ // 更新ID
+ if (toolCall.has("id")) {
+ toolCallInfo.id = toolCall.getString("id")
+ }
+
+ // 更新函数信息
+ if (toolCall.has("function")) {
+ val function = toolCall.getJSONObject("function")
+
+ if (function.has("name")) {
+ toolCallInfo.name = function.getString("name")
+ }
+
+ if (function.has("arguments")) {
+ toolCallInfo.arguments += function.getString("arguments")
+ }
+ }
+ }
+ }
+ }
+ }
+ }
+ } catch (e: Exception) {
+ // 忽略解析错误
+ Log.e(TAG, "解析JSON出错: ${e.message}")
+ }
+ }
+ }
+ } catch (e: Exception) {
+ callback.onError(OpenAIException("处理响应流时出错: ${e.message}"))
+ } finally {
+ responseBody.close()
+ }
+ }
+ })
+ }
+
+ /**
+ * 发送函数调用结果
+ */
+ fun sendFunctionCallResult(
+ messages: JSONArray,
+ functionCall: JSONObject,
+ functionResult: String,
+ callback: StreamCallback
+ ) {
+ try {
+ val fullMessages = JSONArray()
+
+ // 添加用户消息
+ for (i in 0 until messages.length()) {
+ fullMessages.put(messages.getJSONObject(i))
+ }
+
+ // 添加函数调用消息
+ fullMessages.put(JSONObject().apply {
+ put("role", "assistant")
+ put("content", "")
+
+ // 添加工具调用
+ val toolCalls = JSONArray().apply {
+ val toolCall = JSONObject().apply {
+ put("id", functionCall.optString("id", "call_${System.currentTimeMillis()}"))
+ put("type", "function")
+ put("function", JSONObject().apply {
+ put("name", functionCall.getString("name"))
+ put("arguments", functionCall.getString("arguments"))
+ })
+ }
+ put(toolCall)
+ }
+ put("tool_calls", toolCalls)
+ })
+
+ // 添加函数调用结果
+ fullMessages.put(JSONObject().apply {
+ put("role", "tool")
+ put("content", functionResult)
+ put("tool_call_id", functionCall.optString("id", "call_${System.currentTimeMillis()}"))
+ })
+
+ // 添加一个带有content的assistant消息，确保API请求不会因为缺少content而失败
+ fullMessages.put(JSONObject().apply {
+ put("role", "assistant")
+ put("content", "") // 空内容，让模型生成新的回复
+ })
+
+ // 发送完整对话
+ sendMessageStream(fullMessages, callback)
+
+ } catch (e: Exception) {
+ callback.onError(OpenAIException("发送函数调用结果失败: ${e.message}"))
+ }
+ }
+
+ /**
+ * 处理工具调用（处理完整的函数调用并回调）
+ */
+ private fun processToolCalls(toolCalls: Map, callback: StreamCallback) {
+ if (toolCalls.isEmpty()) return
+
+ // 只处理第一个工具调用
+ val firstToolCall = toolCalls.entries.firstOrNull()?.value ?: return
+
+ if (firstToolCall.isValid()) {
+ // 创建函数调用JSON对象
+ val functionCall = JSONObject().apply {
+ put("name", firstToolCall.name)
+ put("arguments", firstToolCall.arguments)
+ put("id", firstToolCall.id)
+ }
+
+ // 回调
+ callback.onFunctionCall(functionCall)
+ }
+ }
+
+ /**
+ * 工具调用信息类
+ */
+ private class ToolCallInfo {
+ var id: String = ""
+ var name: String = ""
+ var arguments: String = ""
+
+ fun isValid(): Boolean {
+ return id.isNotEmpty() && name.isNotEmpty()
+ }
+ }
+
+ /**
+ * 流式输出回调接口
+ */
+ interface StreamCallback {
+ fun onToken(token: String)
+ fun onComplete()
+ fun onError(e: Exception)
+ fun onFunctionCall(functionCall: JSONObject)
+ }
+}
+
+/**
+ * OpenAI服务异常
+ */
+class OpenAIException(message: String) : Exception(message)
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/android/src/main/kotlin/com/yunqiinnovation/open_ai_service/OpenAIServicePlugin.kt b/local_plugins/open_ai_service/android/src/main/kotlin/com/yunqiinnovation/open_ai_service/OpenAIServicePlugin.kt
new file mode 100644
index 000000000..70af36c13
--- /dev/null
+++ b/local_plugins/open_ai_service/android/src/main/kotlin/com/yunqiinnovation/open_ai_service/OpenAIServicePlugin.kt
@@ -0,0 +1,317 @@
+package com.yunqiinnovation.open_ai_service
+
+import android.content.Context
+import android.util.Log
+import androidx.annotation.NonNull
+import io.flutter.embedding.engine.plugins.FlutterPlugin
+import io.flutter.plugin.common.MethodCall
+import io.flutter.plugin.common.MethodChannel
+import io.flutter.plugin.common.MethodChannel.MethodCallHandler
+import io.flutter.plugin.common.MethodChannel.Result
+import io.flutter.plugin.common.EventChannel
+import io.flutter.plugin.common.EventChannel.EventSink
+import io.flutter.plugin.common.EventChannel.StreamHandler
+import org.json.JSONArray
+import org.json.JSONObject
+import java.util.concurrent.CountDownLatch
+import java.util.concurrent.Executors
+
+/** OpenAIServicePlugin */
+class OpenAIServicePlugin : FlutterPlugin, MethodCallHandler, StreamHandler {
+ /// 方法通道名称
+ private val methodChannelName = "com.yunqiinnovation.open_ai_service/methods"
+
+ /// 事件通道名称
+ private val eventChannelName = "com.yunqiinnovation.open_ai_service/events"
+
+ /// 方法通道
+ private lateinit var methodChannel: MethodChannel
+
+ /// 事件通道
+ private lateinit var eventChannel: EventChannel
+
+ /// 应用上下文
+ private lateinit var context: Context
+
+ /// OpenAI服务实例
+ private val openAIService = OpenAIService()
+
+ /// 事件接收器（用于流式输出）
+ private var eventSink: EventSink? = null
+
+ /// 执行器（用于后台线程）
+ private val executor = Executors.newSingleThreadExecutor()
+
+ override fun onAttachedToEngine(@NonNull flutterPluginBinding: FlutterPlugin.FlutterPluginBinding) {
+ // 保存上下文
+ context = flutterPluginBinding.applicationContext
+
+ // 初始化方法通道
+ methodChannel = MethodChannel(flutterPluginBinding.binaryMessenger, methodChannelName)
+ methodChannel.setMethodCallHandler(this)
+
+ // 初始化事件通道
+ eventChannel = EventChannel(flutterPluginBinding.binaryMessenger, eventChannelName)
+ eventChannel.setStreamHandler(this)
+ }
+
+ override fun onMethodCall(@NonNull call: MethodCall, @NonNull result: Result) {
+ when (call.method) {
+ "initialize" -> {
+ val apiKey = call.argument("apiKey") ?: ""
+ val baseUrl = call.argument("baseUrl") ?: ""
+ val model = call.argument("model") ?: ""
+
+ val initialized = openAIService.initialize(apiKey, baseUrl, model)
+ result.success(initialized)
+ }
+
+ "registerFunction" -> {
+ val name = call.argument("name") ?: ""
+ val description = call.argument("description") ?: ""
+ val parameters = call.argument>("parameters")
+
+ if (name.isEmpty() || parameters == null) {
+ result.error("INVALID_ARGUMENT", "函数注册参数无效", null)
+ return
+ }
+
+ val parametersJson = JSONObject(parameters)
+ val registered = openAIService.registerFunction(name, description, parametersJson)
+ result.success(registered)
+ }
+
+ "sendMessage" -> {
+ val messagesRaw = call.argument>>("messages") ?: emptyList()
+
+ // 转换消息格式
+ val messages = JSONArray()
+ for (message in messagesRaw) {
+ messages.put(JSONObject(message))
+ }
+
+ // 在后台线程执行请求
+ executor.execute {
+ try {
+ val response = openAIService.sendMessage(messages)
+ // 在主线程返回结果
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ result.success(response)
+ }
+ } catch (e: Exception) {
+ // 在主线程返回错误
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ result.error("OPENAI_ERROR", e.message, null)
+ }
+ }
+ }
+ }
+
+ "sendMessageStream" -> {
+ val messagesRaw = call.argument>>("messages") ?: emptyList()
+
+ // 检查事件接收器
+ if (eventSink == null) {
+ result.error("NO_EVENT_SINK", "没有可用的事件流接收器", null)
+ return
+ }
+
+ // 转换消息格式
+ val messages = JSONArray()
+ for (message in messagesRaw) {
+ messages.put(JSONObject(message))
+ }
+
+ // 在后台线程执行请求
+ executor.execute {
+ try {
+ openAIService.sendMessageStream(
+ messages = messages,
+ callback = object : OpenAIService.StreamCallback {
+ override fun onToken(token: String) {
+ // 发送token事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ eventSink?.success(mapOf("type" to "token", "content" to token))
+ }
+ }
+
+ override fun onComplete() {
+ // 发送完成事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ eventSink?.success(mapOf("type" to "complete"))
+ }
+ }
+
+ override fun onError(e: Exception) {
+ // 发送错误事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ eventSink?.success(mapOf("type" to "error", "content" to e.message))
+ }
+ }
+
+ override fun onFunctionCall(functionCall: JSONObject) {
+ // 发送函数调用事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ val functionCallMap = functionCall.toMap()
+ eventSink?.success(mapOf("type" to "functionCall", "content" to functionCallMap))
+ }
+ }
+ }
+ )
+
+ // 请求已开始
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ result.success(true)
+ }
+ } catch (e: Exception) {
+ // 在主线程返回错误
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ result.error("OPENAI_ERROR", e.message, null)
+ }
+ }
+ }
+ }
+
+ "sendFunctionCallResult" -> {
+ val messagesRaw = call.argument>>("messages") ?: emptyList()
+ val functionCallRaw = call.argument>("functionCall") ?: emptyMap()
+ val functionResult = call.argument("functionResult") ?: ""
+
+ // 检查事件接收器
+ if (eventSink == null) {
+ result.error("NO_EVENT_SINK", "没有可用的事件流接收器", null)
+ return
+ }
+
+ // 转换消息格式
+ val messages = JSONArray()
+ for (message in messagesRaw) {
+ messages.put(JSONObject(message))
+ }
+
+ // 转换函数调用
+ val functionCall = JSONObject(functionCallRaw)
+
+ // 在后台线程执行请求
+ executor.execute {
+ try {
+ openAIService.sendFunctionCallResult(
+ messages = messages,
+ functionCall = functionCall,
+ functionResult = functionResult,
+ callback = object : OpenAIService.StreamCallback {
+ override fun onToken(token: String) {
+ // 发送token事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ eventSink?.success(mapOf("type" to "token", "content" to token))
+ }
+ }
+
+ override fun onComplete() {
+ // 发送完成事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ eventSink?.success(mapOf("type" to "complete"))
+ }
+ }
+
+ override fun onError(e: Exception) {
+ // 发送错误事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ eventSink?.success(mapOf("type" to "error", "content" to e.message))
+ }
+ }
+
+ override fun onFunctionCall(nestedFunctionCall: JSONObject) {
+ // 发送函数调用事件
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ val functionCallMap = nestedFunctionCall.toMap()
+ eventSink?.success(mapOf("type" to "functionCall", "content" to functionCallMap))
+ }
+ }
+ }
+ )
+
+ // 请求已开始
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ result.success(true)
+ }
+ } catch (e: Exception) {
+ // 在主线程返回错误
+ android.os.Handler(android.os.Looper.getMainLooper()).post {
+ result.error("OPENAI_ERROR", e.message, null)
+ }
+ }
+ }
+ }
+
+ "createUserMessage" -> {
+ val content = call.argument("content") ?: ""
+ val message = openAIService.createUserMessage(content)
+ result.success(message.toMap())
+ }
+
+ "createAssistantMessage" -> {
+ val content = call.argument("content") ?: ""
+ val message = openAIService.createAssistantMessage(content)
+ result.success(message.toMap())
+ }
+
+ else -> {
+ result.notImplemented()
+ }
+ }
+ }
+
+ override fun onDetachedFromEngine(@NonNull binding: FlutterPlugin.FlutterPluginBinding) {
+ methodChannel.setMethodCallHandler(null)
+ eventChannel.setStreamHandler(null)
+ executor.shutdown()
+ }
+
+ // Stream事件处理
+ override fun onListen(arguments: Any?, eventSink: EventSink?) {
+ this.eventSink = eventSink
+ }
+
+ override fun onCancel(arguments: Any?) {
+ this.eventSink = null
+ }
+
+ // 工具方法：JSONObject转Map
+ private fun JSONObject.toMap(): Map {
+ val map = mutableMapOf()
+ val keys = this.keys()
+ while (keys.hasNext()) {
+ val key = keys.next()
+ var value: Any? = this.opt(key)
+
+ value = when (value) {
+ JSONObject.NULL -> null
+ is JSONObject -> value.toMap()
+ is JSONArray -> value.toList()
+ else -> value
+ }
+
+ map[key] = value
+ }
+ return map
+ }
+
+ // 工具方法：JSONArray转List
+ private fun JSONArray.toList(): List {
+ val list = mutableListOf()
+ for (i in 0 until this.length()) {
+ var value: Any? = this.opt(i)
+
+ value = when (value) {
+ JSONObject.NULL -> null
+ is JSONObject -> value.toMap()
+ is JSONArray -> value.toList()
+ else -> value
+ }
+
+ list.add(value)
+ }
+ return list
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/ios/Classes/OpenAIService.swift b/local_plugins/open_ai_service/ios/Classes/OpenAIService.swift
new file mode 100644
index 000000000..671affd73
--- /dev/null
+++ b/local_plugins/open_ai_service/ios/Classes/OpenAIService.swift
@@ -0,0 +1,473 @@
+import Foundation
+
+/// OpenAI服务异常
+public class OpenAIError: Error {
+ let message: String
+
+ init(_ message: String) {
+ self.message = message
+ }
+}
+
+/// 工具调用信息
+private class ToolCallInfo {
+ var id: String = ""
+ var name: String = ""
+ var arguments: String = ""
+
+ var isValid: Bool {
+ return !id.isEmpty && !name.isEmpty
+ }
+}
+
+/// OpenAI服务iOS原生实现
+public class OpenAIService {
+ private let TAG = "OpenAIService"
+ private var baseUrl = "https://api.openai.com/v1/chat/completions"
+ private var apiKey: String = ""
+ private var isInitialized = false
+ private var model: String = "doubao-1-5-lite-32k-250115" // 默认模型
+
+ // 用于存储注册的函数
+ private var registeredFunctions: [[String: Any]] = []
+
+ // URL会话
+ private let session: URLSession
+
+ public init() {
+ // 创建URL会话配置
+ let config = URLSessionConfiguration.default
+ config.timeoutIntervalForRequest = 30.0
+ config.timeoutIntervalForResource = 30.0
+ session = URLSession(configuration: config)
+ }
+
+ /// 创建用户消息
+ public func createUserMessage(content: String) -> [String: Any] {
+ return ["role": "user", "content": content]
+ }
+
+ /// 创建助手消息
+ public func createAssistantMessage(content: String) -> [String: Any] {
+ return ["role": "assistant", "content": content]
+ }
+
+ /// 初始化OpenAI服务
+ public func initialize(apiKey: String, baseUrl: String = "", model: String = "") -> Bool {
+ self.apiKey = apiKey
+ if !baseUrl.isEmpty {
+ self.baseUrl = baseUrl
+ }
+ if !model.isEmpty {
+ self.model = model
+ }
+ isInitialized = !apiKey.isEmpty
+ return isInitialized
+ }
+
+ /// 注册函数
+ public func registerFunction(name: String, description: String, parameters: [String: Any]) -> Bool {
+ do {
+ let function: [String: Any] = [
+ "name": name,
+ "description": description,
+ "parameters": parameters
+ ]
+
+ // 检查是否已存在相同名称的函数
+ if let existingIndex = registeredFunctions.firstIndex(where: { ($0["name"] as? String) == name }) {
+ // 如果已存在，则替换
+ registeredFunctions[existingIndex] = function
+ } else {
+ // 如果不存在，则添加
+ registeredFunctions.append(function)
+ }
+
+ return true
+ } catch {
+ return false
+ }
+ }
+
+ /// 发送消息（非流式输出）
+ public func sendMessage(messages: [[String: Any]], systemPrompt: String) throws -> String {
+ guard isInitialized, !apiKey.isEmpty else {
+ throw OpenAIError("OpenAI服务未初始化")
+ }
+
+ // 构建完整消息，添加系统提示
+ var fullMessages: [[String: Any]] = [
+ ["role": "system", "content": systemPrompt]
+ ]
+ fullMessages.append(contentsOf: messages)
+
+ // 构建请求体
+ var requestDict: [String: Any] = [
+ "model": model,
+ "messages": fullMessages,
+ "temperature": 0.7,
+ "max_tokens": 2000,
+ "stream": false
+ ]
+
+ // 如果有注册的函数，添加到请求中
+ if !registeredFunctions.isEmpty {
+ var tools: [[String: Any]] = []
+ for function in registeredFunctions {
+ let tool: [String: Any] = [
+ "type": "function",
+ "function": function
+ ]
+ tools.append(tool)
+ }
+ requestDict["tools"] = tools
+ }
+
+ // 将请求数据转换为JSON数据
+ guard let jsonData = try? JSONSerialization.data(withJSONObject: requestDict) else {
+ throw OpenAIError("无法序列化请求数据")
+ }
+
+ // 创建URL请求
+ guard let url = URL(string: baseUrl) else {
+ throw OpenAIError("无效的URL")
+ }
+
+ var request = URLRequest(url: url)
+ request.httpMethod = "POST"
+ request.addValue("application/json", forHTTPHeaderField: "Content-Type")
+ request.addValue("Bearer \(apiKey)", forHTTPHeaderField: "Authorization")
+ request.httpBody = jsonData
+
+ // 创建信号量用于同步请求
+ let semaphore = DispatchSemaphore(value: 0)
+ var responseResult: Result = .failure(OpenAIError("未收到响应"))
+
+ // 执行请求
+ let task = session.dataTask(with: request) { data, response, error in
+ if let error = error {
+ responseResult = .failure(OpenAIError("请求失败: \(error.localizedDescription)"))
+ semaphore.signal()
+ return
+ }
+
+ guard let httpResponse = response as? HTTPURLResponse else {
+ responseResult = .failure(OpenAIError("无效的HTTP响应"))
+ semaphore.signal()
+ return
+ }
+
+ guard httpResponse.statusCode == 200 else {
+ responseResult = .failure(OpenAIError("API调用失败: \(httpResponse.statusCode)"))
+ semaphore.signal()
+ return
+ }
+
+ guard let data = data else {
+ responseResult = .failure(OpenAIError("响应数据为空"))
+ semaphore.signal()
+ return
+ }
+
+ do {
+ // 解析JSON响应
+ guard let jsonResponse = try JSONSerialization.jsonObject(with: data) as? [String: Any] else {
+ responseResult = .failure(OpenAIError("无法解析JSON响应"))
+ semaphore.signal()
+ return
+ }
+
+ // 检查是否有函数调用
+ if let choices = jsonResponse["choices"] as? [[String: Any]], !choices.isEmpty,
+ let choice = choices.first,
+ let message = choice["message"] as? [String: Any] {
+
+ // 检查是否有工具调用
+ if let toolCalls = message["tool_calls"] as? [[String: Any]], !toolCalls.isEmpty,
+ let toolCall = toolCalls.first,
+ let function = toolCall["function"] as? [String: Any],
+ let name = function["name"] as? String,
+ let arguments = function["arguments"] as? String,
+ let id = toolCall["id"] as? String {
+
+ let functionCallDict: [String: Any] = [
+ "name": name,
+ "arguments": arguments,
+ "id": id
+ ]
+
+ // 将函数调用转为JSON字符串
+ if let functionCallData = try? JSONSerialization.data(withJSONObject: functionCallDict),
+ let functionCallString = String(data: functionCallData, encoding: .utf8) {
+ responseResult = .success(functionCallString)
+ semaphore.signal()
+ return
+ }
+ }
+
+ // 如果没有工具调用，返回消息内容
+ if let content = message["content"] as? String {
+ responseResult = .success(content)
+ semaphore.signal()
+ return
+ }
+ }
+
+ responseResult = .failure(OpenAIError("无效的响应格式"))
+ semaphore.signal()
+
+ } catch {
+ responseResult = .failure(OpenAIError("解析响应时出错: \(error.localizedDescription)"))
+ semaphore.signal()
+ }
+ }
+
+ task.resume()
+
+ // 等待响应完成
+ _ = semaphore.wait(timeout: .distantFuture)
+
+ // 返回结果或抛出错误
+ switch responseResult {
+ case .success(let result):
+ return result
+ case .failure(let error):
+ throw error
+ }
+ }
+
+ /// 发送消息（流式输出）
+ public func sendMessageStream(messages: [[String: Any]], systemPrompt: String, callback: @escaping StreamCallback) {
+ guard isInitialized, !apiKey.isEmpty else {
+ callback.onError(OpenAIError("OpenAI服务未初始化"))
+ return
+ }
+
+ // 构建完整消息，添加系统提示
+ var fullMessages: [[String: Any]] = [
+ ["role": "system", "content": systemPrompt]
+ ]
+ fullMessages.append(contentsOf: messages)
+
+ // 构建请求体
+ var requestDict: [String: Any] = [
+ "model": model,
+ "messages": fullMessages,
+ "temperature": 0.7,
+ "max_tokens": 2000,
+ "stream": true
+ ]
+
+ // 如果有注册的函数，添加到请求中
+ if !registeredFunctions.isEmpty {
+ var tools: [[String: Any]] = []
+ for function in registeredFunctions {
+ let tool: [String: Any] = [
+ "type": "function",
+ "function": function
+ ]
+ tools.append(tool)
+ }
+ requestDict["tools"] = tools
+ }
+
+ // 将请求数据转换为JSON数据
+ guard let jsonData = try? JSONSerialization.data(withJSONObject: requestDict) else {
+ callback.onError(OpenAIError("无法序列化请求数据"))
+ return
+ }
+
+ // 创建URL请求
+ guard let url = URL(string: baseUrl) else {
+ callback.onError(OpenAIError("无效的URL"))
+ return
+ }
+
+ var request = URLRequest(url: url)
+ request.httpMethod = "POST"
+ request.addValue("application/json", forHTTPHeaderField: "Content-Type")
+ request.addValue("Bearer \(apiKey)", forHTTPHeaderField: "Authorization")
+ request.addValue("text/event-stream", forHTTPHeaderField: "Accept")
+ request.httpBody = jsonData
+
+ // 用于存储函数调用的各个部分
+ var finalToolCalls: [Int: ToolCallInfo] = [:]
+
+ // 创建数据任务
+ let task = session.dataTask(with: request) { data, response, error in
+ if let error = error {
+ callback.onError(OpenAIError("请求失败: \(error.localizedDescription)"))
+ return
+ }
+
+ guard let httpResponse = response as? HTTPURLResponse else {
+ callback.onError(OpenAIError("无效的HTTP响应"))
+ return
+ }
+
+ guard httpResponse.statusCode == 200 else {
+ callback.onError(OpenAIError("API调用失败: \(httpResponse.statusCode)"))
+ return
+ }
+
+ guard let data = data else {
+ callback.onError(OpenAIError("响应数据为空"))
+ return
+ }
+
+ // 处理SSE数据流
+ if let text = String(data: data, encoding: .utf8) {
+ let lines = text.components(separatedBy: "\n")
+
+ for line in lines {
+ if line.isEmpty { continue }
+
+ if line.hasPrefix("data: ") {
+ let dataContent = line.dropFirst(6)
+
+ // 处理[DONE]消息
+ if dataContent == "[DONE]" {
+ self.processToolCalls(finalToolCalls, callback: callback)
+ callback.onComplete()
+ break
+ }
+
+ // 解析JSON数据
+ do {
+ if let data = dataContent.data(using: .utf8),
+ let jsonData = try JSONSerialization.jsonObject(with: data) as? [String: Any] {
+
+ // 处理消息内容
+ if let choices = jsonData["choices"] as? [[String: Any]], !choices.isEmpty,
+ let choice = choices.first {
+
+ if let delta = choice["delta"] as? [String: Any] {
+ // 处理普通文本内容
+ if let content = delta["content"] as? String {
+ callback.onToken(content)
+ }
+
+ // 处理工具调用（函数调用）
+ if let toolCalls = delta["tool_calls"] as? [[String: Any]] {
+ for toolCall in toolCalls {
+ if let index = toolCall["index"] as? Int {
+ // 创建或获取现有的工具调用信息
+ let toolCallInfo = finalToolCalls[index] ?? ToolCallInfo()
+
+ // 更新ID
+ if let id = toolCall["id"] as? String {
+ toolCallInfo.id = id
+ }
+
+ // 更新函数信息
+ if let function = toolCall["function"] as? [String: Any] {
+ if let name = function["name"] as? String {
+ toolCallInfo.name = name
+ }
+
+ if let arguments = function["arguments"] as? String {
+ toolCallInfo.arguments += arguments
+ }
+ }
+
+ finalToolCalls[index] = toolCallInfo
+ }
+ }
+ }
+ }
+ }
+ }
+ } catch {
+ NSLog("解析JSON出错: \(error.localizedDescription)")
+ // 忽略解析错误，继续处理其他行
+ }
+ }
+ }
+ }
+ }
+
+ task.resume()
+ }
+
+ /// 发送函数调用结果
+ public func sendFunctionCallResult(
+ messages: [[String: Any]],
+ systemPrompt: String,
+ functionCall: [String: Any],
+ functionResult: String,
+ callback: @escaping StreamCallback
+ ) {
+ do {
+ // 构建完整消息数组
+ var fullMessages: [[String: Any]] = [
+ // 添加系统提示
+ ["role": "system", "content": systemPrompt]
+ ]
+
+ // 添加用户消息
+ fullMessages.append(contentsOf: messages)
+
+ // 获取函数相关信息
+ guard let name = functionCall["name"] as? String,
+ let arguments = functionCall["arguments"] as? String else {
+ callback.onError(OpenAIError("函数调用信息不完整"))
+ return
+ }
+
+ let id = functionCall["id"] as? String ?? "call_\(Int(Date().timeIntervalSince1970 * 1000))"
+
+ // 添加函数调用消息
+ fullMessages.append([
+ "role": "assistant",
+ "content": NSNull(),
+ "tool_calls": [
+ [
+ "id": id,
+ "type": "function",
+ "function": [
+ "name": name,
+ "arguments": arguments
+ ]
+ ]
+ ]
+ ])
+
+ // 添加函数调用结果
+ fullMessages.append([
+ "role": "tool",
+ "content": functionResult,
+ "tool_call_id": id
+ ])
+
+ // 发送完整对话
+ sendMessageStream(messages: fullMessages, systemPrompt: systemPrompt, callback: callback)
+
+ } catch {
+ callback.onError(OpenAIError("发送函数调用结果失败: \(error.localizedDescription)"))
+ }
+ }
+
+ /// 处理工具调用（函数调用）并回调
+ private func processToolCalls(_ toolCalls: [Int: ToolCallInfo], callback: StreamCallback) {
+ if toolCalls.isEmpty { return }
+
+ // 只处理第一个工具调用
+ guard let firstToolCall = toolCalls.values.first, firstToolCall.isValid else { return }
+
+ // 创建函数调用字典
+ let functionCall: [String: Any] = [
+ "name": firstToolCall.name,
+ "arguments": firstToolCall.arguments,
+ "id": firstToolCall.id
+ ]
+
+ // 回调
+ callback.onFunctionCall(functionCall)
+ }
+
+ /// 流式输出回调协议
+ public typealias StreamCallback = (onToken: (String) -> Void,
+ onComplete: () -> Void,
+ onError: (Error) -> Void,
+ onFunctionCall: ([String: Any]) -> Void)
+}
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/ios/Classes/OpenAIServicePlugin.swift b/local_plugins/open_ai_service/ios/Classes/OpenAIServicePlugin.swift
new file mode 100644
index 000000000..d919e2584
--- /dev/null
+++ b/local_plugins/open_ai_service/ios/Classes/OpenAIServicePlugin.swift
@@ -0,0 +1,215 @@
+import Flutter
+import UIKit
+
+public class OpenAIServicePlugin: NSObject, FlutterPlugin, FlutterStreamHandler {
+ // OpenAI服务实例
+ private let openAIService = OpenAIService()
+
+ // 事件接收器
+ private var eventSink: FlutterEventSink?
+
+ // 注册插件
+ public static func register(with registrar: FlutterPluginRegistrar) {
+ let methodChannel = FlutterMethodChannel(name: "com.yunqiinnovation.open_ai_service/methods", binaryMessenger: registrar.messenger())
+ let eventChannel = FlutterEventChannel(name: "com.yunqiinnovation.open_ai_service/events", binaryMessenger: registrar.messenger())
+
+ let instance = OpenAIServicePlugin()
+ registrar.addMethodCallDelegate(instance, channel: methodChannel)
+ eventChannel.setStreamHandler(instance)
+ }
+
+ // 处理方法调用
+ public func handle(_ call: FlutterMethodCall, result: @escaping FlutterResult) {
+ switch call.method {
+ case "initialize":
+ if let args = call.arguments as? [String: Any],
+ let apiKey = args["apiKey"] as? String {
+ let baseUrl = args["baseUrl"] as? String ?? ""
+ let model = args["model"] as? String ?? ""
+ let initialized = openAIService.initialize(apiKey: apiKey, baseUrl: baseUrl, model: model)
+ result(initialized)
+ } else {
+ result(FlutterError(code: "INVALID_ARGUMENT", message: "初始化参数无效", details: nil))
+ }
+
+ case "registerFunction":
+ if let args = call.arguments as? [String: Any],
+ let name = args["name"] as? String,
+ let description = args["description"] as? String,
+ let parameters = args["parameters"] as? [String: Any] {
+
+ let registered = openAIService.registerFunction(name: name, description: description, parameters: parameters)
+ result(registered)
+ } else {
+ result(FlutterError(code: "INVALID_ARGUMENT", message: "函数注册参数无效", details: nil))
+ }
+
+ case "sendMessage":
+ guard let args = call.arguments as? [String: Any],
+ let messagesRaw = args["messages"] as? [[String: Any]],
+ let systemPrompt = args["systemPrompt"] as? String else {
+ result(FlutterError(code: "INVALID_ARGUMENT", message: "发送消息参数无效", details: nil))
+ return
+ }
+
+ // 在后台线程执行
+ DispatchQueue.global(qos: .userInitiated).async {
+ do {
+ let response = try self.openAIService.sendMessage(messages: messagesRaw, systemPrompt: systemPrompt)
+ // 在主线程返回结果
+ DispatchQueue.main.async {
+ result(response)
+ }
+ } catch {
+ // 在主线程返回错误
+ DispatchQueue.main.async {
+ result(FlutterError(code: "OPENAI_ERROR", message: error.localizedDescription, details: nil))
+ }
+ }
+ }
+
+ case "sendMessageStream":
+ guard let args = call.arguments as? [String: Any],
+ let messagesRaw = args["messages"] as? [[String: Any]],
+ let systemPrompt = args["systemPrompt"] as? String else {
+ result(FlutterError(code: "INVALID_ARGUMENT", message: "发送消息参数无效", details: nil))
+ return
+ }
+
+ // 检查事件接收器
+ guard let eventSink = self.eventSink else {
+ result(FlutterError(code: "NO_EVENT_SINK", message: "没有可用的事件流接收器", details: nil))
+ return
+ }
+
+ // 在后台线程执行
+ DispatchQueue.global(qos: .userInitiated).async {
+ let callback: OpenAIService.StreamCallback = (
+ onToken: { token in
+ // 发送token事件
+ DispatchQueue.main.async {
+ eventSink(["type": "token", "content": token])
+ }
+ },
+ onComplete: {
+ // 发送完成事件
+ DispatchQueue.main.async {
+ eventSink(["type": "complete"])
+ }
+ },
+ onError: { error in
+ // 发送错误事件
+ DispatchQueue.main.async {
+ eventSink(["type": "error", "content": error.localizedDescription])
+ }
+ },
+ onFunctionCall: { functionCall in
+ // 发送函数调用事件
+ DispatchQueue.main.async {
+ eventSink(["type": "functionCall", "content": functionCall])
+ }
+ }
+ )
+
+ self.openAIService.sendMessageStream(messages: messagesRaw, systemPrompt: systemPrompt, callback: callback)
+
+ // 请求已开始
+ DispatchQueue.main.async {
+ result(true)
+ }
+ }
+
+ case "sendFunctionCallResult":
+ guard let args = call.arguments as? [String: Any],
+ let messagesRaw = args["messages"] as? [[String: Any]],
+ let systemPrompt = args["systemPrompt"] as? String,
+ let functionCallRaw = args["functionCall"] as? [String: Any],
+ let functionResult = args["functionResult"] as? String else {
+ result(FlutterError(code: "INVALID_ARGUMENT", message: "发送函数调用结果参数无效", details: nil))
+ return
+ }
+
+ // 检查事件接收器
+ guard let eventSink = self.eventSink else {
+ result(FlutterError(code: "NO_EVENT_SINK", message: "没有可用的事件流接收器", details: nil))
+ return
+ }
+
+ // 在后台线程执行
+ DispatchQueue.global(qos: .userInitiated).async {
+ let callback: OpenAIService.StreamCallback = (
+ onToken: { token in
+ // 发送token事件
+ DispatchQueue.main.async {
+ eventSink(["type": "token", "content": token])
+ }
+ },
+ onComplete: {
+ // 发送完成事件
+ DispatchQueue.main.async {
+ eventSink(["type": "complete"])
+ }
+ },
+ onError: { error in
+ // 发送错误事件
+ DispatchQueue.main.async {
+ eventSink(["type": "error", "content": error.localizedDescription])
+ }
+ },
+ onFunctionCall: { functionCall in
+ // 发送函数调用事件
+ DispatchQueue.main.async {
+ eventSink(["type": "functionCall", "content": functionCall])
+ }
+ }
+ )
+
+ self.openAIService.sendFunctionCallResult(
+ messages: messagesRaw,
+ systemPrompt: systemPrompt,
+ functionCall: functionCallRaw,
+ functionResult: functionResult,
+ callback: callback
+ )
+
+ // 请求已开始
+ DispatchQueue.main.async {
+ result(true)
+ }
+ }
+
+ case "createUserMessage":
+ if let args = call.arguments as? [String: Any],
+ let content = args["content"] as? String {
+ let message = openAIService.createUserMessage(content: content)
+ result(message)
+ } else {
+ result(FlutterError(code: "INVALID_ARGUMENT", message: "创建用户消息参数无效", details: nil))
+ }
+
+ case "createAssistantMessage":
+ if let args = call.arguments as? [String: Any],
+ let content = args["content"] as? String {
+ let message = openAIService.createAssistantMessage(content: content)
+ result(message)
+ } else {
+ result(FlutterError(code: "INVALID_ARGUMENT", message: "创建助手消息参数无效", details: nil))
+ }
+
+ default:
+ result(FlutterMethodNotImplemented)
+ }
+ }
+
+ // MARK: - FlutterStreamHandler
+
+ public func onListen(withArguments arguments: Any?, eventSink events: @escaping FlutterEventSink) -> FlutterError? {
+ self.eventSink = events
+ return nil
+ }
+
+ public func onCancel(withArguments arguments: Any?) -> FlutterError? {
+ self.eventSink = nil
+ return nil
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/lib/open_ai_service.dart b/local_plugins/open_ai_service/lib/open_ai_service.dart
new file mode 100644
index 000000000..7c6fe8024
--- /dev/null
+++ b/local_plugins/open_ai_service/lib/open_ai_service.dart
@@ -0,0 +1,267 @@
+import 'dart:async';
+import 'dart:convert';
+
+import 'package:flutter/services.dart';
+
+/// OpenAI服务异常
+class OpenAIException implements Exception {
+ final String message;
+
+ OpenAIException(this.message);
+
+ @override
+ String toString() => 'OpenAIException: $message';
+}
+
+/// OpenAI服务事件类型
+enum OpenAIEventType {
+ token,
+ complete,
+ error,
+ functionCall,
+}
+
+/// OpenAI服务事件
+class OpenAIEvent {
+ final OpenAIEventType type;
+ final dynamic content;
+
+ OpenAIEvent({required this.type, this.content});
+
+ factory OpenAIEvent.fromMap(Map map) {
+ final typeStr = map['type'] as String;
+ final content = map['content'];
+
+ return OpenAIEvent(
+ type: _typeFromString(typeStr),
+ content: content,
+ );
+ }
+
+ static OpenAIEventType _typeFromString(String typeStr) {
+ switch (typeStr) {
+ case 'token':
+ return OpenAIEventType.token;
+ case 'complete':
+ return OpenAIEventType.complete;
+ case 'error':
+ return OpenAIEventType.error;
+ case 'functionCall':
+ return OpenAIEventType.functionCall;
+ default:
+ throw ArgumentError('未知的事件类型: $typeStr');
+ }
+ }
+}
+
+/// OpenAI服务插件
+class OpenAIService {
+ static const MethodChannel _channel = MethodChannel('com.yunqiinnovation.open_ai_service/methods');
+ static const EventChannel _eventChannel = EventChannel('com.yunqiinnovation.open_ai_service/events');
+
+ /// 事件流控制器
+ StreamController? _eventStreamController;
+
+ /// 事件流
+ Stream? _eventStream;
+
+ /// 获取事件流
+ Stream get eventStream {
+ if (_eventStream == null) {
+ _eventStreamController = StreamController.broadcast();
+ _eventStream = _eventStreamController!.stream;
+
+ // 监听原生事件
+ _eventChannel.receiveBroadcastStream().listen(
+ (dynamic event) {
+ if (event is Map) {
+ final eventMap = Map.from(event);
+ final openAIEvent = OpenAIEvent.fromMap(eventMap);
+ _eventStreamController!.add(openAIEvent);
+ }
+ },
+ onError: (error) {
+ _eventStreamController!.addError(OpenAIException('事件流错误: $error'));
+ },
+ );
+ }
+
+ return _eventStream!;
+ }
+
+ /// 初始化OpenAI服务
+ ///
+ /// [apiKey] OpenAI API密钥
+ /// [baseUrl] 可选，自定义API基础URL
+ /// [model] 可选，自定义使用的模型
+ Future initialize({
+ required String apiKey,
+ String baseUrl = '',
+ String model = '',
+ }) async {
+ try {
+ final result = await _channel.invokeMethod(
+ 'initialize',
+ {
+ 'apiKey': apiKey,
+ 'baseUrl': baseUrl,
+ 'model': model,
+ },
+ );
+
+ return result ?? false;
+ } catch (e) {
+ throw OpenAIException('初始化失败: $e');
+ }
+ }
+
+ /// 注册函数
+ ///
+ /// [name] 函数名称
+ /// [description] 函数描述
+ /// [parameters] 函数参数
+ Future registerFunction({
+ required String name,
+ required String description,
+ required Map parameters,
+ }) async {
+ try {
+ final result = await _channel.invokeMethod(
+ 'registerFunction',
+ {
+ 'name': name,
+ 'description': description,
+ 'parameters': parameters,
+ },
+ );
+
+ return result ?? false;
+ } catch (e) {
+ throw OpenAIException('注册函数失败: $e');
+ }
+ }
+
+ /// 创建用户消息
+ ///
+ /// [content] 消息内容
+ Future> createUserMessage(String content) async {
+ try {
+ final result = await _channel.invokeMethod>(
+ 'createUserMessage',
+ {'content': content},
+ );
+
+ if (result == null) {
+ throw OpenAIException('创建用户消息失败: 结果为空');
+ }
+
+ return Map.from(result);
+ } catch (e) {
+ throw OpenAIException('创建用户消息失败: $e');
+ }
+ }
+
+ /// 创建助手消息
+ ///
+ /// [content] 消息内容
+ Future> createAssistantMessage(String content) async {
+ try {
+ final result = await _channel.invokeMethod>(
+ 'createAssistantMessage',
+ {'content': content},
+ );
+
+ if (result == null) {
+ throw OpenAIException('创建助手消息失败: 结果为空');
+ }
+
+ return Map.from(result);
+ } catch (e) {
+ throw OpenAIException('创建助手消息失败: $e');
+ }
+ }
+
+ /// 发送消息（非流式输出）
+ ///
+ /// [messages] 消息列表
+ Future sendMessage({
+ required List> messages,
+ }) async {
+ try {
+ final result = await _channel.invokeMethod(
+ 'sendMessage',
+ {
+ 'messages': messages,
+ },
+ );
+
+ if (result == null) {
+ throw OpenAIException('发送消息失败: 结果为空');
+ }
+
+ return result;
+ } catch (e) {
+ throw OpenAIException('发送消息失败: $e');
+ }
+ }
+
+ /// 发送消息（流式输出）
+ ///
+ /// [messages] 消息列表
+ ///
+ /// 返回一个布尔值，表示请求是否已开始
+ Future sendMessageStream({
+ required List> messages,
+ }) async {
+ try {
+ final result = await _channel.invokeMethod(
+ 'sendMessageStream',
+ {
+ 'messages': messages,
+ },
+ );
+
+ return result ?? false;
+ } catch (e) {
+ throw OpenAIException('发送流式消息失败: $e');
+ }
+ }
+
+ /// 发送函数调用结果
+ ///
+ /// [messages] 消息列表
+ /// [functionCall] 函数调用信息
+ /// [functionResult] 函数调用结果
+ ///
+ /// 返回一个布尔值，表示请求是否已开始
+ Future sendFunctionCallResult({
+ required List> messages,
+ required Map functionCall,
+ required String functionResult,
+ }) async {
+ try {
+ final result = await _channel.invokeMethod(
+ 'sendFunctionCallResult',
+ {
+ 'messages': messages,
+ 'functionCall': functionCall,
+ 'functionResult': functionResult,
+ },
+ );
+
+ return result ?? false;
+ } catch (e) {
+ throw OpenAIException('发送函数调用结果失败: $e');
+ }
+ }
+
+
+ /// 从JSON字符串解析函数调用
+ Map parseFunctionCall(String functionCallJson) {
+ try {
+ return json.decode(functionCallJson) as Map;
+ } catch (e) {
+ throw OpenAIException('解析函数调用失败: $e');
+ }
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/open_ai_service/pubspec.yaml b/local_plugins/open_ai_service/pubspec.yaml
new file mode 100644
index 000000000..8e00ed3ea
--- /dev/null
+++ b/local_plugins/open_ai_service/pubspec.yaml
@@ -0,0 +1,26 @@
+name: open_ai_service
+description: 原生OpenAI服务插件，提供与OpenAI API的交互功能，支持流式输出和函数调用。
+version: 0.0.1
+homepage: https://github.com/yunqiinnovation/deep_voice
+
+environment:
+ sdk: ">=2.17.0 <4.0.0"
+ flutter: ">=2.5.0"
+
+dependencies:
+ flutter:
+ sdk: flutter
+
+dev_dependencies:
+ flutter_test:
+ sdk: flutter
+ flutter_lints: ^2.0.0
+
+flutter:
+ plugin:
+ platforms:
+ android:
+ package: com.yunqiinnovation.open_ai_service
+ pluginClass: OpenAIServicePlugin
+ ios:
+ pluginClass: OpenAIServicePlugin
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/README.md b/local_plugins/volcano_speech/README.md
new file mode 100644
index 000000000..572317b57
--- /dev/null
+++ b/local_plugins/volcano_speech/README.md
@@ -0,0 +1,206 @@
+# 火山引擎语音服务插件
+
+基于火山引擎语音SDK封装的Flutter插件，支持Android平台。
+
+## 功能
+- 语音合成（TTS）
+ - 支持在线/离线/混合合成模式
+ - 支持SSML格式文本
+ - 支持情感合成和情感预测
+ - 支持音量、语速、音高等多种参数调节
+ - 复刻音色支持
+ - 离线资源管理
+- 语音识别（ASR）
+ - 一次性识别
+ - 连续识别
+ - 长按识别
+ - 热词优化
+ - 多语言支持
+
+## 使用说明
+
+### TTS 语音合成
+
+#### 基础用法
+```dart
+import 'package:volcano_speech/volcano_speech.dart';
+
+// 初始化
+await VolcanoSpeech.initialize(appId: 'YOUR_APP_ID', apiKey: 'YOUR_API_KEY');
+
+// 设置音量、语速等参数
+await VolcanoSpeech.setSpeechParams(
+ rate: 0, // 语速: -500到500，0为正常速度
+ volume: 100, // 音量: 0到100，默认100
+ pitch: 0, // 音高: -500到500，0为正常音高
+ silenceDuration: 500, // 静音段时长: 毫秒
+);
+
+// 设置音色
+await VolcanoSpeech.setVoice(
+ voiceName: 'zh_female_qingxin',
+ voiceType: 'qingxin'
+);
+
+// 开始合成并播放
+await VolcanoSpeech.speakText(text: '这是一段测试文本');
+
+// 暂停播放
+await VolcanoSpeech.pausePlayback();
+
+// 继续播放
+await VolcanoSpeech.resumePlayback();
+
+// 停止播放
+await VolcanoSpeech.stopSpeaking();
+
+// 销毁引擎
+await VolcanoSpeech.dispose();
+```
+
+#### 进阶用法
+```dart
+// 设置工作模式
+await VolcanoSpeech.setWorkMode(mode: TtsWorkMode.alternate); // 先在线，断网时切换离线
+
+// 设置离线发音人
+await VolcanoSpeech.setOfflineVoice(
+ voiceName: 'xifei',
+ voiceType: 'qingxin'
+);
+
+// 下载离线资源
+await VolcanoSpeech.downloadOfflineResource(
+ voiceTypes: ['qingxin', 'zhenjiang'],
+ languages: ['zh-CN']
+);
+
+// 使用SSML格式文本
+await VolcanoSpeech.setTextType(type: TtsTextType.ssml);
+await VolcanoSpeech.speakText(
+ text: '这是一段2023-10-01的语音合成'
+);
+
+// 设置情感
+await VolcanoSpeech.setEmotion(emotion: 'happy');
+
+// 启用情感预测
+await VolcanoSpeech.setEnableEmotionPredict(enable: true);
+
+// 启用服务端缓存
+await VolcanoSpeech.setEnableCache(enable: true);
+
+// 监听TTS进度事件
+VolcanoSpeech.ttsProgressEvents.listen((event) {
+ print('播放进度: ${(event.progress * 100).toStringAsFixed(1)}%');
+});
+
+// 复刻音色支持
+await VolcanoSpeech.setEnableVoiceClone(
+ enable: true,
+ backendCluster: 'your_cluster_name'
+);
+```
+
+### ASR 语音识别
+#### 一次性识别
+```dart
+import 'package:volcano_speech/volcano_speech.dart';
+
+// 初始化
+await VolcanoSpeech.initializeAsr(
+ appId: 'YOUR_APP_ID',
+ apiKey: 'YOUR_API_KEY',
+ supportedLanguages: ['zh-CN'],
+);
+
+// 设置识别语言
+await VolcanoSpeech.setAsrLanguage(language: 'zh-CN');
+
+// 设置热词（可选）
+await VolcanoSpeech.setAsrHotWords(
+ hotWords: '{"hotwords":[{"word":"火山引擎","scale":2.0}]}',
+);
+
+// 开始一次性识别
+try {
+ final result = await VolcanoSpeech.recognizeOnce();
+ print('识别结果: ${result['text']}, 语言: ${result['language']}');
+} catch (e) {
+ print('识别出错: $e');
+}
+
+// 销毁引擎
+await VolcanoSpeech.disposeAsr();
+```
+
+#### 连续识别
+```dart
+import 'package:volcano_speech/volcano_speech.dart';
+
+// 初始化
+await VolcanoSpeech.initializeAsr(
+ appId: 'YOUR_APP_ID',
+ apiKey: 'YOUR_API_KEY',
+);
+
+// 监听识别事件
+VolcanoSpeech.asrEvents.listen((event) {
+ switch (event.type) {
+ case AsrEventType.sessionStarted:
+ print('识别会话开始');
+ break;
+ case AsrEventType.sessionStopped:
+ print('识别会话结束');
+ break;
+ case AsrEventType.recognizing:
+ print('正在识别: ${event.text}');
+ break;
+ case AsrEventType.result:
+ print('识别结果: ${event.text}');
+ break;
+ case AsrEventType.volumeChanged:
+ print('音量: ${event.volume}');
+ break;
+ case AsrEventType.error:
+ print('识别错误: ${event.errorMessage}');
+ break;
+ }
+});
+
+// 开始连续识别
+await VolcanoSpeech.startContinuousRecognition();
+
+// 检查是否正在识别
+final isActive = await VolcanoSpeech.isContinuousRecognitionActive();
+print('是否正在识别: $isActive');
+
+// 停止连续识别
+await VolcanoSpeech.stopContinuousRecognition();
+```
+
+#### 长按识别
+```dart
+import 'package:volcano_speech/volcano_speech.dart';
+
+// 初始化
+await VolcanoSpeech.initializeAsr(
+ appId: 'YOUR_APP_ID',
+ apiKey: 'YOUR_API_KEY',
+);
+
+// 监听识别事件
+VolcanoSpeech.asrEvents.listen((event) {
+ // 处理事件...
+});
+
+// 用户按下按钮时开始识别
+onPressed: () async {
+ await VolcanoSpeech.startListening();
+},
+
+// 用户释放按钮时停止识别
+onReleased: () async {
+ await VolcanoSpeech.stopListening();
+},
+```
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/android/build.gradle.kts b/local_plugins/volcano_speech/android/build.gradle.kts
new file mode 100644
index 000000000..86b03fd0e
--- /dev/null
+++ b/local_plugins/volcano_speech/android/build.gradle.kts
@@ -0,0 +1,64 @@
+import com.android.build.gradle.LibraryExtension
+
+buildscript {
+ repositories {
+ google()
+ mavenCentral()
+ }
+ dependencies {
+ classpath("com.android.tools.build:gradle:7.3.0")
+ classpath("org.jetbrains.kotlin:kotlin-gradle-plugin:1.7.10")
+ }
+}
+
+allprojects {
+ repositories {
+ google()
+ mavenCentral()
+
+ }
+}
+
+plugins {
+ id("com.android.library")
+ kotlin("android")
+}
+
+// 配置android扩展
+configure {
+ namespace = "com.yunqiinnovation.volcano_speech"
+ compileSdkVersion(33)
+
+ defaultConfig {
+ minSdk = 21
+ }
+
+ compileOptions {
+ sourceCompatibility = JavaVersion.VERSION_11
+ targetCompatibility = JavaVersion.VERSION_11
+ }
+
+ sourceSets {
+ getByName("main") {
+ manifest.srcFile("src/main/AndroidManifest.xml")
+ java.srcDirs("src/main/kotlin")
+ }
+ }
+
+ // 添加lint选项
+ lintOptions {
+ isCheckReleaseBuilds = false
+ }
+}
+
+// 显式设置Kotlin JVM目标版本
+tasks.withType {
+ kotlinOptions {
+ jvmTarget = "11"
+ }
+}
+
+dependencies {
+ // 添加火山引擎语音合成SDK
+ implementation("com.bytedance.speechengine:speechengine_tob:0.0.5")
+}
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/android/settings.gradle.kts b/local_plugins/volcano_speech/android/settings.gradle.kts
new file mode 100644
index 000000000..7f5342866
--- /dev/null
+++ b/local_plugins/volcano_speech/android/settings.gradle.kts
@@ -0,0 +1 @@
+rootProject.name = "volcano_speech"
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/android/src/main/AndroidManifest.xml b/local_plugins/volcano_speech/android/src/main/AndroidManifest.xml
new file mode 100644
index 000000000..338e8293c
--- /dev/null
+++ b/local_plugins/volcano_speech/android/src/main/AndroidManifest.xml
@@ -0,0 +1,9 @@
+
+
+
+
+
+
+
+
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoAsrHelper.kt b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoAsrHelper.kt
new file mode 100644
index 000000000..614474d1a
--- /dev/null
+++ b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoAsrHelper.kt
@@ -0,0 +1,529 @@
+package com.yunqiinnovation.volcano_speech
+
+import android.content.Context
+import android.os.Handler
+import android.os.Looper
+import com.bytedance.speech.speechengine.SpeechEngine
+import com.bytedance.speech.speechengine.SpeechEngineDefines
+import com.bytedance.speech.speechengine.SpeechEngineGenerator
+import com.yunqiinnovation.volcano_speech.utils.FileLogger
+import org.json.JSONArray
+import org.json.JSONException
+import org.json.JSONObject
+
+/**
+ * 火山语音识别帮助类 (大模型版本)
+ */
+class VolcanoAsrHelper(private val context: Context) {
+ private val TAG = "VolcanoAsrHelper"
+ private val mainHandler = Handler(Looper.getMainLooper())
+
+ // 语音引擎相关
+ private var engine: SpeechEngine? = null
+ private var engineHandler: Long = -1
+ private var isInitialized = false
+
+ // 当前回调
+ private var currentAsrCallback: ASRCallback? = null
+ private var currentContinuousCallback: ASRContinuousCallback? = null
+
+ // 识别状态
+ private var isContinuousRecognitionActive = false
+
+ // 配置参数
+ private var language = "zh-CN"
+ private var enableVolume = false
+ private var showUtterances = false
+
+ /**
+ * ASR一次性识别回调
+ */
+ interface ASRCallback {
+ fun onSuccess(text: String)
+ fun onError(error: String)
+ }
+
+ /**
+ * ASR连续识别回调
+ */
+ interface ASRContinuousCallback {
+ fun onResult(text: String)
+ fun onRecognizing(text: String)
+ fun onSessionStarted()
+ fun onSessionStopped()
+ fun onVolumeChanged(volume: Int)
+ fun onError(error: String)
+ }
+
+ /**
+ * 初始化语音识别引擎
+ */
+ fun initialize(appId: String, token: String, resourceId: String): Boolean {
+ if (isInitialized) {
+ FileLogger.i(TAG, "引擎已经初始化")
+ return true
+ }
+
+ try {
+ // 准备环境
+ SpeechEngineGenerator.PrepareEnvironment(context, null)
+
+ // 创建引擎
+ engine = SpeechEngineGenerator.getInstance()
+ engineHandler = engine?.createEngine() ?: -1
+
+ if (engineHandler == -1L) {
+ FileLogger.e(TAG, "创建引擎失败")
+ return false
+ }
+
+ // 设置上下文
+ engine?.setContext(context)
+
+ // 设置引擎类型为ASR
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ENGINE_NAME_STRING, SpeechEngineDefines.ASR_ENGINE)
+
+ // 设置日志级别
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_LOG_LEVEL_STRING, SpeechEngineDefines.LOG_LEVEL_WARN)
+
+ // 设置用户ID和设备ID (使用静态值，实际项目中应替换为真实值)
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_UID_STRING, "user_id")
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_DEVICE_ID_STRING, "device_id")
+
+ // 设置鉴权信息 - 大模型版本不需要Bearer前缀
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_APP_ID_STRING, appId)
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_APP_TOKEN_STRING, token)
+
+ // 设置资源ID - 大模型必需
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_RESOURCE_ID_STRING, resourceId)
+
+ // 设置协议类型为Seed - 大模型必需
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_PROTOCOL_TYPE_INT, SpeechEngineDefines.PROTOCOL_TYPE_SEED)
+
+ // 设置网络配置
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_ADDRESS_STRING, "wss://openspeech.bytedance.com")
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_URI_STRING, "/api/v3/sauc/bigmodel")
+
+ // 设置超时时间
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_CONN_TIMEOUT_INT, 12000)
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_RECV_TIMEOUT_INT, 8000)
+
+ // 设置音频来源为录音机
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_RECORDER_TYPE_STRING, SpeechEngineDefines.RECORDER_TYPE_RECORDER)
+
+ // 设置最大录音时长 (默认60秒)
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_VAD_MAX_SPEECH_DURATION_INT, 60000)
+
+ // 设置回声消除 (用于ASR识别时不会收到TTS的声音)
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_RECORDER_PRESET_INT, SpeechEngineDefines.RECORDER_PRESET_VOICE_COMMUNICATION)
+
+ // 初始化引擎
+ val result = engine?.initEngine(engineHandler)
+ isInitialized = result == SpeechEngineDefines.ERR_NO_ERROR
+
+ if (isInitialized) {
+ FileLogger.i(TAG, "引擎初始化成功")
+
+ // 设置回调监听
+ engine?.setListener(object : SpeechEngine.SpeechListener {
+ override fun onSpeechMessage(type: Int, data: ByteArray, len: Int) {
+ val stdData = String(data)
+ handleEngineEvent(type, stdData)
+ }
+ })
+
+ return true
+ } else {
+ FileLogger.e(TAG, "引擎初始化失败: $result")
+ return false
+ }
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "初始化异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置语言
+ */
+ fun setLanguage(language: String): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ this.language = language
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_LANGUAGE_STRING, language)
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置语言失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置热词
+ */
+ fun setHotWords(hotWordsId: String): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ if (hotWordsId.isNotEmpty()) {
+ // 大模型ASR需要通过请求参数设置热词
+ val reqParams = "{\"corpus\":{\"boosting_table_id\":\"$hotWordsId\"}}"
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_REQ_PARAMS_STRING, reqParams)
+ }
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置热词失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置是否返回音量
+ */
+ fun setEnableVolume(enable: Boolean): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ this.enableVolume = enable
+ engine?.setOptionBoolean(engineHandler, SpeechEngineDefines.PARAMS_KEY_ENABLE_GET_VOLUME_BOOL, enable)
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置音量返回失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置是否显示语音停顿、分句、分词信息
+ */
+ fun setShowUtterances(show: Boolean): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ this.showUtterances = show
+ engine?.setOptionBoolean(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_SHOW_UTTER_BOOL, show)
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置语音信息显示失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置VAD切句参数
+ */
+ fun setVadParams(forceToSpeechTime: Int, endWindowSize: Int): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ val reqParams = "{\"force_to_speech_time\":$forceToSpeechTime, \"end_window_size\":$endWindowSize}"
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_REQ_PARAMS_STRING, reqParams)
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置VAD参数失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置纠错词表
+ */
+ fun setCorrectWords(correctWordsJson: String): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ val reqParams = "{\"context\": \"{\\\"correct_words\\\": $correctWordsJson}\"}"
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ASR_REQ_PARAMS_STRING, reqParams)
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置纠错词表失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 一次性识别
+ */
+ fun recognizeOnce(callback: ASRCallback): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ if (isContinuousRecognitionActive) {
+ FileLogger.e(TAG, "当前正在连续识别中")
+ return false
+ }
+
+ try {
+ this.currentAsrCallback = callback
+
+ // 停止当前引擎
+ engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_SYNC_STOP_ENGINE, "")
+
+ // 启动引擎开始识别
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_START_ENGINE, "")
+
+ if (ret != SpeechEngineDefines.ERR_NO_ERROR) {
+ FileLogger.e(TAG, "启动识别失败: $ret")
+ return false
+ }
+
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "识别异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 停止一次性识别
+ */
+ fun stopRecognize(): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ // 告知引擎音频输入完成
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_FINISH_TALKING, "")
+
+ if (ret != SpeechEngineDefines.ERR_NO_ERROR) {
+ FileLogger.e(TAG, "停止识别失败: $ret")
+ return false
+ }
+
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "停止识别异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 开始连续识别
+ */
+ fun startContinuousRecognition(callback: ASRContinuousCallback): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ if (isContinuousRecognitionActive) {
+ FileLogger.e(TAG, "当前已经在连续识别中")
+ return false
+ }
+
+ try {
+ this.currentContinuousCallback = callback
+
+ // 停止当前引擎
+ engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_SYNC_STOP_ENGINE, "")
+
+ // 启动引擎开始识别
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_START_ENGINE, "")
+
+ if (ret != SpeechEngineDefines.ERR_NO_ERROR) {
+ FileLogger.e(TAG, "启动连续识别失败: $ret")
+ return false
+ }
+
+ isContinuousRecognitionActive = true
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "连续识别异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 停止连续识别
+ */
+ fun stopContinuousRecognition(): Boolean {
+ if (!isInitialized || !isContinuousRecognitionActive) {
+ FileLogger.e(TAG, "引擎未初始化或未在连续识别中")
+ return false
+ }
+
+ try {
+ // 停止引擎
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_STOP_ENGINE, "")
+
+ if (ret != SpeechEngineDefines.ERR_NO_ERROR) {
+ FileLogger.e(TAG, "停止连续识别失败: $ret")
+ return false
+ }
+
+ isContinuousRecognitionActive = false
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "停止连续识别异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 判断是否在连续识别中
+ */
+ fun isContinuousRecognitionActive(): Boolean {
+ return isContinuousRecognitionActive
+ }
+
+ /**
+ * 释放资源
+ */
+ fun release() {
+ if (!isInitialized) {
+ return
+ }
+
+ try {
+ // 停止引擎
+ if (isContinuousRecognitionActive) {
+ stopContinuousRecognition()
+ } else {
+ engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_STOP_ENGINE, "")
+ }
+
+ // 销毁引擎
+ engine?.destroyEngine(engineHandler)
+ engineHandler = -1
+ engine = null
+ isInitialized = false
+
+ FileLogger.i(TAG, "引擎已释放")
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "释放引擎异常: ${e.message}", e)
+ }
+ }
+
+ /**
+ * 处理引擎事件
+ */
+ private fun handleEngineEvent(type: Int, data: String) {
+ when (type) {
+ SpeechEngineDefines.MESSAGE_TYPE_ENGINE_START -> {
+ // 引擎启动成功
+ mainHandler.post {
+ currentContinuousCallback?.onSessionStarted()
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_ENGINE_STOP -> {
+ // 引擎停止
+ isContinuousRecognitionActive = false
+ mainHandler.post {
+ currentContinuousCallback?.onSessionStopped()
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_PARTIAL_RESULT -> {
+ // 中间识别结果
+ try {
+ val json = JSONObject(data)
+ if (!json.has("result")) {
+ return
+ }
+
+ val resultArray = json.getJSONArray("result")
+ if (resultArray.length() > 0) {
+ val result = resultArray.getJSONObject(0)
+ val text = result.optString("text", "")
+
+ if (text.isNotEmpty()) {
+ mainHandler.post {
+ currentContinuousCallback?.onRecognizing(text)
+ }
+ }
+ }
+ } catch (e: JSONException) {
+ FileLogger.e(TAG, "解析中间识别结果异常: ${e.message}", e)
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_FINAL_RESULT -> {
+ // 最终识别结果
+ try {
+ val json = JSONObject(data)
+ if (!json.has("result")) {
+ return
+ }
+
+ val resultArray = json.getJSONArray("result")
+ if (resultArray.length() > 0) {
+ val result = resultArray.getJSONObject(0)
+ val text = result.optString("text", "")
+
+ if (text.isNotEmpty()) {
+ mainHandler.post {
+ if (isContinuousRecognitionActive) {
+ currentContinuousCallback?.onResult(text)
+ } else {
+ currentAsrCallback?.onSuccess(text)
+ }
+ }
+ }
+ }
+ } catch (e: JSONException) {
+ FileLogger.e(TAG, "解析最终识别结果异常: ${e.message}", e)
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_VOLUME_LEVEL -> {
+ // 音量回调
+ if (enableVolume && isContinuousRecognitionActive) {
+ try {
+ val volume = (data.toFloat() * 100).toInt()
+ mainHandler.post {
+ currentContinuousCallback?.onVolumeChanged(volume)
+ }
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "解析音量数据异常: ${e.message}", e)
+ }
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_ENGINE_ERROR -> {
+ // 错误信息
+ try {
+ val json = JSONObject(data)
+ val errCode = json.optInt("err_code", -1)
+ val errMsg = json.optString("err_msg", "未知错误")
+
+ FileLogger.e(TAG, "引擎错误: $errCode, $errMsg")
+
+ mainHandler.post {
+ if (isContinuousRecognitionActive) {
+ currentContinuousCallback?.onError(errMsg)
+ isContinuousRecognitionActive = false
+ } else {
+ currentAsrCallback?.onError(errMsg)
+ }
+ }
+ } catch (e: JSONException) {
+ FileLogger.e(TAG, "解析错误信息异常: ${e.message}", e)
+ }
+ }
+ }
+ }
+}
diff --git a/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoSpeechPlugin.kt b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoSpeechPlugin.kt
new file mode 100644
index 000000000..bb6f1e452
--- /dev/null
+++ b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoSpeechPlugin.kt
@@ -0,0 +1,445 @@
+package com.yunqiinnovation.volcano_speech
+
+import android.content.Context
+import android.os.Handler
+import android.os.Looper
+import androidx.annotation.NonNull
+import com.yunqiinnovation.volcano_speech.utils.FileLogger
+
+import io.flutter.embedding.engine.plugins.FlutterPlugin
+import io.flutter.plugin.common.MethodCall
+import io.flutter.plugin.common.MethodChannel
+import io.flutter.plugin.common.MethodChannel.MethodCallHandler
+import io.flutter.plugin.common.MethodChannel.Result
+import io.flutter.plugin.common.EventChannel
+
+/** VolcanoSpeechPlugin */
+class VolcanoSpeechPlugin: FlutterPlugin {
+ private val TAG = "VolcanoSpeechPlugin"
+ private lateinit var context: Context
+ private val mainHandler = Handler(Looper.getMainLooper())
+
+ // ASR相关
+ private lateinit var asrChannel: MethodChannel
+ private lateinit var asrEventChannel: EventChannel
+ private var asrEventSink: EventChannel.EventSink? = null
+
+ // TTS相关
+ private lateinit var ttsChannel: MethodChannel
+ private lateinit var ttsEventChannel: EventChannel
+ private var ttsEventSink: EventChannel.EventSink? = null
+
+ // TTS帮助类
+ private lateinit var volcanoTtsHelper: VolcanoTtsHelper
+
+ // ASR帮助类
+ private lateinit var volcanoAsrHelper: VolcanoAsrHelper
+
+ override fun onAttachedToEngine(@NonNull flutterPluginBinding: FlutterPlugin.FlutterPluginBinding) {
+ context = flutterPluginBinding.applicationContext
+
+ // 初始化ASR通道
+ asrChannel = MethodChannel(flutterPluginBinding.binaryMessenger, "volcano_speech/asr")
+ asrChannel.setMethodCallHandler(AsrMethodHandler())
+
+ // 初始化TTS通道
+ ttsChannel = MethodChannel(flutterPluginBinding.binaryMessenger, "volcano_speech/tts")
+ ttsChannel.setMethodCallHandler(TtsMethodHandler())
+
+ // 初始化ASR事件通道
+ asrEventChannel = EventChannel(flutterPluginBinding.binaryMessenger, "volcano_speech/asr_events")
+ asrEventChannel.setStreamHandler(object : EventChannel.StreamHandler {
+ override fun onListen(arguments: Any?, events: EventChannel.EventSink?) {
+ asrEventSink = events
+ }
+
+ override fun onCancel(arguments: Any?) {
+ asrEventSink = null
+ }
+ })
+
+ // 初始化TTS事件通道
+ ttsEventChannel = EventChannel(flutterPluginBinding.binaryMessenger, "volcano_speech/tts_events")
+ ttsEventChannel.setStreamHandler(object : EventChannel.StreamHandler {
+ override fun onListen(arguments: Any?, events: EventChannel.EventSink?) {
+ ttsEventSink = events
+ }
+
+ override fun onCancel(arguments: Any?) {
+ ttsEventSink = null
+ }
+ })
+
+ // 初始化TTS帮助类
+ volcanoTtsHelper = VolcanoTtsHelper(context)
+
+ // 初始化ASR帮助类
+ volcanoAsrHelper = VolcanoAsrHelper(context)
+ }
+
+ // 发送ASR事件
+ private fun sendAsrEvent(event: Map) {
+ FileLogger.d(TAG, "发送ASR事件: $event")
+ if (asrEventSink == null) {
+ FileLogger.w(TAG, "无法发送ASR事件：事件通道未准备好")
+ return
+ }
+
+ mainHandler.post {
+ try {
+ asrEventSink?.success(event)
+ FileLogger.d(TAG, "ASR事件发送成功")
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "发送ASR事件失败: ${e.message}")
+ }
+ }
+ }
+
+ // 发送TTS事件
+ private fun sendTtsEvent(event: Map) {
+ FileLogger.d(TAG, "发送TTS事件: $event")
+ if (ttsEventSink == null) {
+ FileLogger.w(TAG, "无法发送TTS事件：事件通道未准备好")
+ return
+ }
+
+ mainHandler.post {
+ try {
+ ttsEventSink?.success(event)
+ FileLogger.d(TAG, "TTS事件发送成功")
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "发送TTS事件失败: ${e.message}")
+ }
+ }
+ }
+
+ // ASR方法处理器
+ inner class AsrMethodHandler : MethodCallHandler {
+ override fun onMethodCall(@NonNull call: MethodCall, @NonNull result: Result) {
+ when (call.method) {
+ "getPlatformVersion" -> {
+ result.success("Android ${android.os.Build.VERSION.RELEASE}")
+ }
+ "initialize" -> {
+ val appId = call.argument("appId") ?: ""
+ val apiKey = call.argument("apiKey") ?: ""
+ val resourceId = call.argument("resourceId") ?: ""
+
+ if (appId.isEmpty() || apiKey.isEmpty() || resourceId.isEmpty()) {
+ result.error("INVALID_ARGUMENTS", "appId、apiKey和resourceId不能为空", null)
+ return
+ }
+
+ val success = volcanoAsrHelper.initialize(appId, apiKey, resourceId)
+ result.success(success)
+ }
+ "setLanguage" -> {
+ val language = call.argument("language") ?: "zh-CN"
+ result.success(volcanoAsrHelper.setLanguage(language))
+ }
+ "setHotWords" -> {
+ val hotWordsId = call.argument("hotWordsId") ?: ""
+ result.success(volcanoAsrHelper.setHotWords(hotWordsId))
+ }
+ "setVadParams" -> {
+ val forceToSpeechTime = call.argument("forceToSpeechTime") ?: 0
+ val endWindowSize = call.argument("endWindowSize") ?: 800
+ result.success(volcanoAsrHelper.setVadParams(forceToSpeechTime, endWindowSize))
+ }
+ "setCorrectWords" -> {
+ val correctWordsJson = call.argument("correctWordsJson") ?: "{}"
+ result.success(volcanoAsrHelper.setCorrectWords(correctWordsJson))
+ }
+ "setEnableVolume" -> {
+ val enable = call.argument("enable") ?: false
+ result.success(volcanoAsrHelper.setEnableVolume(enable))
+ }
+ "setShowUtterances" -> {
+ val enable = call.argument("enable") ?: false
+ result.success(volcanoAsrHelper.setShowUtterances(enable))
+ }
+ "recognizeOnce" -> {
+ // 确保当前不在连续识别中
+ if (volcanoAsrHelper.isContinuousRecognitionActive()) {
+ result.error("ASR_BUSY", "当前正在连续识别中", null)
+ return
+ }
+
+ volcanoAsrHelper.recognizeOnce(object : VolcanoAsrHelper.ASRCallback {
+ override fun onSuccess(text: String) {
+ mainHandler.post {
+ result.success(mapOf(
+ "text" to text
+ ))
+ }
+ }
+
+ override fun onError(error: String) {
+ mainHandler.post {
+ result.error("ASR_ERROR", error, null)
+ }
+ }
+ })
+ }
+ "startContinuousRecognition" -> {
+ // 确保事件通道已准备好
+ if (asrEventSink == null) {
+ result.error("EVENT_CHANNEL_NOT_READY", "事件通道未准备好，无法开始连续识别", null)
+ return
+ }
+
+ val success = volcanoAsrHelper.startContinuousRecognition(object : VolcanoAsrHelper.ASRContinuousCallback {
+ override fun onResult(text: String) {
+ sendAsrEvent(mapOf(
+ "type" to "result",
+ "text" to text
+ ))
+ }
+
+ override fun onRecognizing(text: String) {
+ sendAsrEvent(mapOf(
+ "type" to "recognizing",
+ "text" to text
+ ))
+ }
+
+ override fun onSessionStarted() {
+ sendAsrEvent(mapOf("type" to "sessionStarted"))
+ }
+
+ override fun onSessionStopped() {
+ sendAsrEvent(mapOf("type" to "sessionStopped"))
+ }
+
+ override fun onVolumeChanged(volume: Int) {
+ sendAsrEvent(mapOf(
+ "type" to "volumeChanged",
+ "volume" to volume
+ ))
+ }
+
+ override fun onError(error: String) {
+ sendAsrEvent(mapOf(
+ "type" to "error",
+ "message" to error
+ ))
+ }
+ })
+
+ result.success(success)
+ }
+ "stopContinuousRecognition" -> {
+ val success = volcanoAsrHelper.stopContinuousRecognition()
+ result.success(success)
+ }
+ "stopRecognize" -> {
+ val success = volcanoAsrHelper.stopRecognize()
+ result.success(success)
+ }
+ "isContinuousRecognitionActive" -> {
+ result.success(volcanoAsrHelper.isContinuousRecognitionActive())
+ }
+ "release" -> {
+ volcanoAsrHelper.release()
+ result.success(true)
+ }
+ else -> {
+ result.notImplemented()
+ }
+ }
+ }
+ }
+
+ // TTS方法处理器
+ inner class TtsMethodHandler : MethodCallHandler {
+ override fun onMethodCall(@NonNull call: MethodCall, @NonNull result: Result) {
+ when (call.method) {
+ "getPlatformVersion" -> {
+ result.success("Android ${android.os.Build.VERSION.RELEASE}")
+ }
+ "initialize" -> {
+ val appId = call.argument("appId") ?: ""
+ val token = call.argument("token") ?: ""
+ val resourceId = call.argument("resourceId") ?: ""
+
+ if (appId.isEmpty() || token.isEmpty() || resourceId.isEmpty()) {
+ result.error("INVALID_ARGUMENTS", "appId、token和resourceId不能为空", null)
+ return
+ }
+
+ val success = volcanoTtsHelper.initialize(appId, token, resourceId)
+ result.success(success)
+ }
+ "setVoice" -> {
+ val voice = call.argument("voice") ?: ""
+
+ if (voice.isEmpty()) {
+ result.error("INVALID_ARGUMENTS", "voice不能为空", null)
+ return
+ }
+
+ result.success(volcanoTtsHelper.setVoice(voice))
+ }
+ "setContinuousMode" -> {
+ val isContinuous = call.argument("isContinuous") ?: false
+ result.success(volcanoTtsHelper.setContinuousMode(isContinuous))
+ }
+ "speak" -> {
+ val text = call.argument("text") ?: ""
+
+ if (text.isEmpty()) {
+ result.error("INVALID_ARGUMENTS", "合成文本不能为空", null)
+ return
+ }
+
+ volcanoTtsHelper.speak(text, object : VolcanoTtsHelper.TTSCallback {
+ override fun onStart(reqId: String) {
+ sendTtsEvent(mapOf(
+ "type" to "start",
+ "reqId" to reqId
+ ))
+ }
+
+ override fun onProgress(reqId: String, progress: Double) {
+ sendTtsEvent(mapOf(
+ "type" to "progress",
+ "reqId" to reqId,
+ "progress" to progress
+ ))
+ }
+
+ override fun onComplete(reqId: String) {
+ sendTtsEvent(mapOf(
+ "type" to "complete",
+ "reqId" to reqId
+ ))
+
+ mainHandler.post {
+ result.success(true)
+ }
+ }
+
+ override fun onError(reqId: String, errorCode: Int, errorMsg: String) {
+ sendTtsEvent(mapOf(
+ "type" to "error",
+ "reqId" to reqId,
+ "errorCode" to errorCode,
+ "errorMsg" to errorMsg
+ ))
+
+ mainHandler.post {
+ result.error("TTS_ERROR", errorMsg, null)
+ }
+ }
+ })
+ }
+ "synthesisNext" -> {
+ val text = call.argument("text") ?: ""
+
+ if (text.isEmpty()) {
+ result.error("INVALID_ARGUMENTS", "合成文本不能为空", null)
+ return
+ }
+
+ val success = volcanoTtsHelper.synthesisNext(text)
+ result.success(success)
+ }
+ "pause" -> {
+ result.success(volcanoTtsHelper.pause())
+ }
+ "resume" -> {
+ result.success(volcanoTtsHelper.resume())
+ }
+ "stop" -> {
+ result.success(volcanoTtsHelper.stop())
+ }
+ "release" -> {
+ volcanoTtsHelper.release()
+ result.success(true)
+ }
+ // 以下方法用于与Azure Speech版本兼容
+ "setSpeechSynthesisVoice" -> {
+ val voiceName = call.argument("voiceName") ?: ""
+
+ if (voiceName.isEmpty()) {
+ result.error("INVALID_ARGUMENTS", "语音名称不能为空", null)
+ return
+ }
+
+ result.success(volcanoTtsHelper.setVoice(voiceName))
+ }
+ "speakText" -> {
+ val text = call.argument("text") ?: ""
+
+ if (text.isEmpty()) {
+ result.error("INVALID_ARGUMENTS", "合成文本不能为空", null)
+ return
+ }
+
+ volcanoTtsHelper.speak(text, object : VolcanoTtsHelper.TTSCallback {
+ override fun onStart(reqId: String) {
+ sendTtsEvent(mapOf(
+ "type" to "start",
+ "reqId" to reqId
+ ))
+ }
+
+ override fun onProgress(reqId: String, progress: Double) {
+ sendTtsEvent(mapOf(
+ "type" to "progress",
+ "reqId" to reqId,
+ "progress" to progress
+ ))
+ }
+
+ override fun onComplete(reqId: String) {
+ sendTtsEvent(mapOf(
+ "type" to "complete",
+ "reqId" to reqId
+ ))
+
+ mainHandler.post {
+ result.success(true)
+ }
+ }
+
+ override fun onError(reqId: String, errorCode: Int, errorMsg: String) {
+ sendTtsEvent(mapOf(
+ "type" to "error",
+ "reqId" to reqId,
+ "errorCode" to errorCode,
+ "errorMsg" to errorMsg
+ ))
+
+ mainHandler.post {
+ result.error("TTS_ERROR", errorMsg, null)
+ }
+ }
+ })
+ }
+ "stopSpeaking" -> {
+ result.success(volcanoTtsHelper.stop())
+ }
+ "pauseSpeaking" -> {
+ result.success(volcanoTtsHelper.pause())
+ }
+ "resumeSpeaking" -> {
+ result.success(volcanoTtsHelper.resume())
+ }
+ else -> {
+ result.notImplemented()
+ }
+ }
+ }
+ }
+
+ override fun onDetachedFromEngine(@NonNull binding: FlutterPlugin.FlutterPluginBinding) {
+ asrChannel.setMethodCallHandler(null)
+ ttsChannel.setMethodCallHandler(null)
+ asrEventChannel.setStreamHandler(null)
+ ttsEventChannel.setStreamHandler(null)
+
+ volcanoTtsHelper.release()
+ volcanoAsrHelper.release()
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoTtsHelper.kt b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoTtsHelper.kt
new file mode 100644
index 000000000..e24a1ceb1
--- /dev/null
+++ b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/VolcanoTtsHelper.kt
@@ -0,0 +1,399 @@
+package com.yunqiinnovation.volcano_speech
+
+import android.content.Context
+import android.os.Handler
+import android.os.Looper
+import com.bytedance.speech.speechengine.SpeechEngine
+import com.bytedance.speech.speechengine.SpeechEngineDefines
+import com.bytedance.speech.speechengine.SpeechEngineGenerator
+import com.yunqiinnovation.volcano_speech.utils.FileLogger
+import org.json.JSONObject
+
+/**
+ * 火山语音合成帮助类 (大模型版本)
+ */
+class VolcanoTtsHelper(private val context: Context) {
+ private val TAG = "VolcanoTtsHelper"
+ private val mainHandler = Handler(Looper.getMainLooper())
+
+ // 语音引擎相关
+ private var engine: SpeechEngine? = null
+ private var engineHandler: Long = -1
+ private var isInitialized = false
+
+ // 当前回调
+ private var currentTtsCallback: TTSCallback? = null
+
+ // 合成状态
+ private var isPlaying = false
+ private var currentReqId = ""
+
+ // 配置参数
+ private var voice = "zh_female_yuxi"
+ private var ttsText = ""
+ private var isContinuous = false
+
+ /**
+ * TTS合成回调接口
+ */
+ interface TTSCallback {
+ fun onStart(reqId: String)
+ fun onProgress(reqId: String, progress: Double)
+ fun onComplete(reqId: String)
+ fun onError(reqId: String, errorCode: Int, errorMsg: String)
+ }
+
+ /**
+ * 初始化语音合成引擎
+ */
+ fun initialize(appId: String, token: String, resourceId: String): Boolean {
+ if (isInitialized) {
+ FileLogger.i(TAG, "引擎已经初始化")
+ return true
+ }
+
+ try {
+ // 准备环境
+ SpeechEngineGenerator.PrepareEnvironment(context, null)
+
+ // 创建引擎
+ engine = SpeechEngineGenerator.getInstance()
+ engineHandler = engine?.createEngine() ?: -1
+
+ if (engineHandler == -1L) {
+ FileLogger.e(TAG, "创建引擎失败")
+ return false
+ }
+
+ // 设置上下文
+ engine?.setContext(context)
+
+ // 设置引擎类型为TTS
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_ENGINE_NAME_STRING, SpeechEngineDefines.TTS_ENGINE)
+
+ // 设置日志级别
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_LOG_LEVEL_STRING, SpeechEngineDefines.LOG_LEVEL_WARN)
+
+ // 设置用户ID和设备ID (使用静态值，实际项目中应替换为真实值)
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_UID_STRING, "user_id")
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_DEVICE_ID_STRING, "device_id")
+
+ // 设置授权信息 - 大模型版本不需要Bearer前缀
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_APP_ID_STRING, appId)
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_APP_TOKEN_STRING, token)
+
+ // 设置资源ID - 大模型必需
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_RESOURCE_ID_STRING, resourceId)
+
+ // 设置协议类型为Seed - 大模型必需
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_PROTOCOL_TYPE_INT, SpeechEngineDefines.PROTOCOL_TYPE_SEED)
+
+ // 设置合成策略为在线合成
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_WORK_MODE_INT, SpeechEngineDefines.TTS_WORK_MODE_ONLINE)
+
+ // 设置在线请求资源配置
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_ADDRESS_STRING, "wss://openspeech.bytedance.com")
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_URI_STRING, "/api/v3/sauc/bigmodel")
+
+ // 设置发音人
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_VOICE_ONLINE_STRING, voice)
+
+ // 设置播放进度回调
+ engine?.setOptionInt(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_WITH_FRONTEND_INT, 1)
+
+ // 初始化引擎
+ val result = engine?.initEngine(engineHandler)
+ isInitialized = result == SpeechEngineDefines.ERR_NO_ERROR
+
+ if (isInitialized) {
+ FileLogger.i(TAG, "引擎初始化成功")
+
+ // 设置回调监听
+ engine?.setListener(object : SpeechEngine.SpeechListener {
+ override fun onSpeechMessage(type: Int, data: ByteArray, len: Int) {
+ val stdData = String(data)
+ handleEngineEvent(type, stdData)
+ }
+ })
+
+ return true
+ } else {
+ FileLogger.e(TAG, "引擎初始化失败: $result")
+ return false
+ }
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "初始化异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置发音人
+ */
+ fun setVoice(voice: String): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ this.voice = voice
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_VOICE_ONLINE_STRING, voice)
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置发音人失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 设置合成场景（单次或连续）
+ */
+ fun setContinuousMode(isContinuous: Boolean): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ this.isContinuous = isContinuous
+ val scenarioType = if (isContinuous) {
+ SpeechEngineDefines.TTS_SCENARIO_TYPE_NOVEL
+ } else {
+ SpeechEngineDefines.TTS_SCENARIO_TYPE_NORMAL
+ }
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_SCENARIO_STRING, scenarioType)
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "设置合成场景失败: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 开始合成并播放
+ */
+ fun speak(text: String, callback: TTSCallback? = null): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ if (text.isEmpty()) {
+ FileLogger.e(TAG, "合成文本为空")
+ return false
+ }
+
+ try {
+ this.ttsText = text
+ this.currentTtsCallback = callback
+
+ // 设置要合成的文本
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_TEXT_STRING, text)
+
+ // 停止当前引擎
+ engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_SYNC_STOP_ENGINE, "")
+
+ // 启动引擎开始合成
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_START_ENGINE, "")
+
+ if (ret != SpeechEngineDefines.ERR_NO_ERROR) {
+ FileLogger.e(TAG, "启动合成失败: $ret")
+ return false
+ }
+
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "合成异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 仅用于连续合成场景：在引擎启动后添加新的文本进行合成
+ */
+ fun synthesisNext(text: String): Boolean {
+ if (!isInitialized || !isContinuous) {
+ FileLogger.e(TAG, "引擎未初始化或非连续合成模式")
+ return false
+ }
+
+ if (text.isEmpty()) {
+ FileLogger.e(TAG, "合成文本为空")
+ return false
+ }
+
+ try {
+ // 设置要合成的文本
+ engine?.setOptionString(engineHandler, SpeechEngineDefines.PARAMS_KEY_TTS_TEXT_STRING, text)
+
+ // 发送合成指令
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_SYNTHESIS, "")
+
+ if (ret != SpeechEngineDefines.ERR_NO_ERROR) {
+ FileLogger.e(TAG, "添加合成文本失败: $ret")
+ return false
+ }
+
+ return true
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "添加合成文本异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 暂停播放
+ */
+ fun pause(): Boolean {
+ if (!isInitialized || !isPlaying) {
+ FileLogger.e(TAG, "引擎未初始化或未在播放")
+ return false
+ }
+
+ try {
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_PAUSE_PLAYER, "")
+ return ret == SpeechEngineDefines.ERR_NO_ERROR
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "暂停播放异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 恢复播放
+ */
+ fun resume(): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_RESUME_PLAYER, "")
+ return ret == SpeechEngineDefines.ERR_NO_ERROR
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "恢复播放异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 停止播放
+ */
+ fun stop(): Boolean {
+ if (!isInitialized) {
+ FileLogger.e(TAG, "引擎未初始化")
+ return false
+ }
+
+ try {
+ val ret = engine?.sendDirective(engineHandler, SpeechEngineDefines.DIRECTIVE_STOP_ENGINE, "")
+ isPlaying = false
+ return ret == SpeechEngineDefines.ERR_NO_ERROR
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "停止播放异常: ${e.message}", e)
+ return false
+ }
+ }
+
+ /**
+ * 释放资源
+ */
+ fun release() {
+ if (!isInitialized) {
+ return
+ }
+
+ try {
+ // 停止引擎
+ stop()
+
+ // 销毁引擎
+ engine?.destroyEngine(engineHandler)
+ engineHandler = -1
+ engine = null
+ isInitialized = false
+
+ FileLogger.i(TAG, "引擎已释放")
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "释放引擎异常: ${e.message}", e)
+ }
+ }
+
+ /**
+ * 处理引擎事件
+ */
+ private fun handleEngineEvent(type: Int, data: String) {
+ when (type) {
+ SpeechEngineDefines.MESSAGE_TYPE_ENGINE_START -> {
+ // 引擎启动成功，获取请求ID
+ currentReqId = data
+ isPlaying = true
+
+ mainHandler.post {
+ currentTtsCallback?.onStart(currentReqId)
+ }
+
+ if (isContinuous) {
+ // 在连续合成模式下，需要单独发送合成指令
+ synthesisNext(ttsText)
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_ENGINE_STOP -> {
+ // 引擎停止
+ isPlaying = false
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_TTS_START_PLAYING -> {
+ // 开始播放
+ FileLogger.d(TAG, "开始播放: $data")
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_TTS_FINISH_PLAYING -> {
+ // 播放结束
+ isPlaying = false
+
+ mainHandler.post {
+ currentTtsCallback?.onComplete(currentReqId)
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_TTS_PLAYBACK_PROGRESS -> {
+ // 播放进度
+ try {
+ val json = JSONObject(data)
+ val progress = json.optDouble("progress", 0.0)
+ val reqId = json.optString("reqid", "")
+
+ mainHandler.post {
+ currentTtsCallback?.onProgress(reqId, progress)
+ }
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "解析进度信息异常: ${e.message}", e)
+ }
+ }
+
+ SpeechEngineDefines.MESSAGE_TYPE_ENGINE_ERROR -> {
+ // 错误信息
+ try {
+ val json = JSONObject(data)
+ val reqId = json.optString("reqid", "")
+ val errCode = json.optInt("err_code", -1)
+ val errMsg = json.optString("err_msg", "未知错误")
+
+ FileLogger.e(TAG, "引擎错误: $errCode, $errMsg")
+
+ isPlaying = false
+
+ mainHandler.post {
+ currentTtsCallback?.onError(reqId, errCode, errMsg)
+ }
+ } catch (e: Exception) {
+ FileLogger.e(TAG, "解析错误信息异常: ${e.message}", e)
+ }
+ }
+ }
+ }
+}
diff --git a/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/utils/FileLogger.kt b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/utils/FileLogger.kt
new file mode 100644
index 000000000..8fdfb1227
--- /dev/null
+++ b/local_plugins/volcano_speech/android/src/main/kotlin/com/yunqiinnovation/volcano_speech/utils/FileLogger.kt
@@ -0,0 +1,55 @@
+package com.yunqiinnovation.volcano_speech.utils
+
+import android.util.Log
+
+/**
+ * 文件日志记录工具
+ */
+object FileLogger {
+ private const val TAG = "VolcanoSpeech"
+ private var isDebugEnabled = true
+
+ /**
+ * 设置是否启用调试日志
+ */
+ fun setDebugEnabled(enabled: Boolean) {
+ isDebugEnabled = enabled
+ }
+
+ /**
+ * 记录调试日志
+ */
+ fun d(tag: String, message: String) {
+ if (isDebugEnabled) {
+ Log.d("$TAG-$tag", message)
+ }
+ }
+
+ /**
+ * 记录信息日志
+ */
+ fun i(tag: String, message: String) {
+ Log.i("$TAG-$tag", message)
+ }
+
+ /**
+ * 记录警告日志
+ */
+ fun w(tag: String, message: String) {
+ Log.w("$TAG-$tag", message)
+ }
+
+ /**
+ * 记录错误日志
+ */
+ fun e(tag: String, message: String) {
+ Log.e("$TAG-$tag", message)
+ }
+
+ /**
+ * 记录错误日志，带异常
+ */
+ fun e(tag: String, message: String, throwable: Throwable) {
+ Log.e("$TAG-$tag", message, throwable)
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/lib/volcano_speech.dart b/local_plugins/volcano_speech/lib/volcano_speech.dart
new file mode 100644
index 000000000..99da7d4b7
--- /dev/null
+++ b/local_plugins/volcano_speech/lib/volcano_speech.dart
@@ -0,0 +1,603 @@
+import 'dart:async';
+import 'package:flutter/services.dart';
+
+/// ASR 事件类型
+enum AsrEventType {
+ /// 会话开始
+ sessionStarted,
+
+ /// 会话结束
+ sessionStopped,
+
+ /// 正在识别（中间结果）
+ recognizing,
+
+ /// 识别结果（最终结果）
+ result,
+
+ /// 音量变化
+ volumeChanged,
+
+ /// 错误
+ error
+}
+
+/// TTS 工作模式
+enum TtsWorkMode {
+ /// 在线合成
+ online,
+
+ /// 离线合成
+ offline,
+
+ /// 同时在线离线
+ both,
+
+ /// 先在线再离线（网络不好时自动切换）
+ alternate,
+
+ /// 文件模式
+ file
+}
+
+/// TTS 文本类型
+enum TtsTextType {
+ /// 纯文本
+ plain,
+
+ /// SSML格式
+ ssml
+}
+
+/// 协议类型
+enum ProtocolType {
+ /// 默认协议
+ defaultProtocol,
+
+ /// Seed协议（用于大模型）
+ seed
+}
+
+/// TTS 播放进度事件
+class TtsProgressEvent {
+ /// 播放进度 0.0-1.0
+ final double progress;
+
+ /// 请求ID
+ final String reqId;
+
+ const TtsProgressEvent({
+ required this.progress,
+ required this.reqId,
+ });
+
+ @override
+ String toString() {
+ return 'TtsProgressEvent{progress: $progress, reqId: $reqId}';
+ }
+}
+
+/// ASR 事件
+class AsrEvent {
+ /// 事件类型
+ final AsrEventType type;
+
+ /// 识别文本（仅在recognizing和result类型时有效）
+ final String? text;
+
+ /// 识别语言（仅在recognizing和result类型时有效）
+ final String? language;
+
+ /// 音量值（仅在volumeChanged类型时有效）
+ final int? volume;
+
+ /// 错误信息（仅在error类型时有效）
+ final String? errorMessage;
+
+ const AsrEvent({
+ required this.type,
+ this.text,
+ this.language,
+ this.volume,
+ this.errorMessage,
+ });
+
+ factory AsrEvent.fromMap(Map map) {
+ final typeStr = map['type'] as String;
+
+ AsrEventType type;
+ switch (typeStr) {
+ case 'sessionStarted':
+ type = AsrEventType.sessionStarted;
+ break;
+ case 'sessionStopped':
+ type = AsrEventType.sessionStopped;
+ break;
+ case 'recognizing':
+ type = AsrEventType.recognizing;
+ break;
+ case 'result':
+ type = AsrEventType.result;
+ break;
+ case 'volumeChanged':
+ type = AsrEventType.volumeChanged;
+ break;
+ case 'error':
+ type = AsrEventType.error;
+ break;
+ default:
+ throw ArgumentError('未知的事件类型: $typeStr');
+ }
+
+ return AsrEvent(
+ type: type,
+ text: map['text'] as String?,
+ language: map['language'] as String?,
+ volume: map['volume'] as int?,
+ errorMessage: map['message'] as String?,
+ );
+ }
+
+ @override
+ String toString() {
+ return 'AsrEvent{type: $type, text: $text, language: $language, volume: $volume, errorMessage: $errorMessage}';
+ }
+}
+
+/// 火山引擎语音服务插件
+class VolcanoSpeech {
+ static final VolcanoSpeechAsr asr = VolcanoSpeechAsr._();
+ static final VolcanoSpeechTts tts = VolcanoSpeechTts._();
+
+ /// 释放资源
+ static Future dispose() async {
+ await asr.dispose();
+ await tts.dispose();
+ }
+}
+
+/// 火山引擎语音识别服务
+class VolcanoSpeechAsr {
+ static const MethodChannel _channel = MethodChannel('volcano_speech/asr');
+ static const EventChannel _eventChannel = EventChannel('volcano_speech/asr_events');
+
+ /// ASR事件流控制器
+ static final StreamController _eventStreamController = StreamController.broadcast();
+
+ /// ASR事件流
+ Stream get events => _eventStreamController.stream;
+
+ /// 是否已初始化事件监听
+ bool _eventListenerInitialized = false;
+
+ VolcanoSpeechAsr._() {
+ _initEventListener();
+ }
+
+ /// 初始化ASR事件监听
+ void _initEventListener() {
+ if (_eventListenerInitialized) return;
+
+ _eventChannel.receiveBroadcastStream().listen((dynamic event) {
+ if (event is Map) {
+ _eventStreamController.add(AsrEvent.fromMap(event));
+ }
+ });
+
+ _eventListenerInitialized = true;
+ }
+
+ /// 获取平台版本信息
+ Future getPlatformVersion() async {
+ return await _channel.invokeMethod('getPlatformVersion');
+ }
+
+ /// 初始化语音识别引擎
+ ///
+ /// [appId] 火山引擎AppID
+ /// [apiKey] 火山引擎ApiKey
+ /// [supportedLanguages] 支持的语言列表
+ /// [useBigModel] 是否使用大模型识别
+ /// [resourceId] 大模型资源ID（仅当useBigModel为true时有效）
+ Future initialize({
+ required String appId,
+ required String apiKey,
+ List supportedLanguages = const ['zh-CN'],
+ bool useBigModel = false,
+ String resourceId = '',
+ }) async {
+ return await _channel.invokeMethod('initialize', {
+ 'appId': appId,
+ 'apiKey': apiKey,
+ 'supportedLanguages': supportedLanguages,
+ 'useBigModel': useBigModel,
+ 'resourceId': resourceId,
+ }) ?? false;
+ }
+
+ /// 设置识别语言
+ ///
+ /// [language] 语言代码，例如 zh-CN、en-US
+ Future setLanguage(String language) async {
+ return await _channel.invokeMethod('setLanguage', {
+ 'language': language,
+ }) ?? false;
+ }
+
+ /// 设置热词
+ ///
+ /// [hotWords] 热词JSON字符串，例如 {"hotwords":[{"word":"快速入门","scale":2.0}]}
+ Future setHotWords(String hotWords) async {
+ return await _channel.invokeMethod('setHotWords', {
+ 'hotWords': hotWords,
+ }) ?? false;
+ }
+
+ /// 设置ASR请求参数
+ ///
+ /// [params] 请求参数JSON字符串
+ Future setRequestParams(String params) async {
+ return await _channel.invokeMethod('setRequestParams', {
+ 'params': params,
+ }) ?? false;
+ }
+
+ /// 启用语音停顿、分句、分词信息输出
+ ///
+ /// [enable] 是否启用
+ Future setShowUtterances(bool enable) async {
+ return await _channel.invokeMethod('setShowUtterances', {
+ 'enable': enable,
+ }) ?? false;
+ }
+
+ /// 一次性识别（直到说话结束）
+ ///
+ /// 返回识别结果文本和语言
+ Future> recognizeOnce() async {
+ final result = await _channel.invokeMethod('recognizeOnce');
+ return {
+ 'text': result['text'] ?? '',
+ 'language': result['language'] ?? 'zh-CN',
+ };
+ }
+
+ /// 开始连续识别
+ ///
+ /// 通过[events]流监听识别结果
+ Future startContinuousRecognition() async {
+ return await _channel.invokeMethod('startContinuousRecognition') ?? false;
+ }
+
+ /// 停止连续识别
+ Future stopContinuousRecognition() async {
+ return await _channel.invokeMethod('stopContinuousRecognition') ?? false;
+ }
+
+ /// 检查是否正在连续识别
+ Future isContinuousRecognitionActive() async {
+ return await _channel.invokeMethod('isContinuousRecognitionActive') ?? false;
+ }
+
+ /// 开始长按识别（按下开始，抬起结束）
+ ///
+ /// 通过[events]流监听识别结果
+ Future startListening() async {
+ return await _channel.invokeMethod('startListening') ?? false;
+ }
+
+ /// 停止长按识别
+ Future stopListening() async {
+ return await _channel.invokeMethod('stopListening') ?? false;
+ }
+
+ /// 释放ASR资源
+ Future dispose() async {
+ return await _channel.invokeMethod('dispose') ?? false;
+ }
+}
+
+/// 火山引擎语音合成服务
+class VolcanoSpeechTts {
+ static const MethodChannel _channel = MethodChannel('volcano_speech/tts');
+ static const EventChannel _eventChannel = EventChannel('volcano_speech/tts_events');
+
+ /// TTS进度事件流控制器
+ static final StreamController _progressStreamController = StreamController.broadcast();
+
+ /// TTS进度事件流
+ Stream get progressEvents => _progressStreamController.stream;
+
+ /// 是否已初始化事件监听
+ bool _eventListenerInitialized = false;
+
+ VolcanoSpeechTts._() {
+ _initEventListener();
+ }
+
+ /// 初始化TTS事件监听
+ void _initEventListener() {
+ if (_eventListenerInitialized) return;
+
+ _eventChannel.receiveBroadcastStream().listen((dynamic event) {
+ if (event is Map) {
+ final progress = event['progress'] as double?;
+ final reqId = event['reqId'] as String?;
+
+ if (progress != null && reqId != null) {
+ _progressStreamController.add(TtsProgressEvent(
+ progress: progress,
+ reqId: reqId,
+ ));
+ }
+ }
+ });
+
+ _eventListenerInitialized = true;
+ }
+
+ /// 获取平台版本信息
+ Future getPlatformVersion() async {
+ return await _channel.invokeMethod('getPlatformVersion');
+ }
+
+ /// 启用大模型TTS
+ ///
+ /// [enable] 是否启用大模型TTS
+ /// [resourceId] 资源ID
+ Future enableBigModelTts({
+ required bool enable,
+ String resourceId = '',
+ }) async {
+ return await _channel.invokeMethod('enableBigModelTts', {
+ 'enable': enable,
+ 'resourceId': resourceId,
+ }) ?? false;
+ }
+
+ /// 初始化语音合成引擎
+ ///
+ /// [appId] 火山引擎AppID
+ /// [apiKey] 火山引擎ApiKey
+ Future initialize({
+ required String appId,
+ required String apiKey,
+ }) async {
+ return await _channel.invokeMethod('initialize', {
+ 'appId': appId,
+ 'apiKey': apiKey,
+ }) ?? false;
+ }
+
+ /// 设置音色
+ ///
+ /// [voiceName] 音色名称
+ /// [voiceType] 音色类型，默认 qingxin
+ Future setVoice({
+ required String voiceName,
+ String voiceType = 'qingxin',
+ }) async {
+ return await _channel.invokeMethod('setVoice', {
+ 'voiceName': voiceName,
+ 'voiceType': voiceType,
+ }) ?? false;
+ }
+
+ /// 设置离线音色
+ ///
+ /// [voiceName] 音色名称
+ /// [voiceType] 音色类型，默认 qingxin
+ Future setOfflineVoice({
+ required String voiceName,
+ String voiceType = 'qingxin',
+ }) async {
+ return await _channel.invokeMethod('setOfflineVoice', {
+ 'voiceName': voiceName,
+ 'voiceType': voiceType,
+ }) ?? false;
+ }
+
+ /// 设置大模型声音ID
+ ///
+ /// [voiceId] 声音ID
+ Future setBigModelVoiceId(String voiceId) async {
+ return await _channel.invokeMethod('setBigModelVoiceId', {
+ 'voiceId': voiceId,
+ }) ?? false;
+ }
+
+ /// 设置大模型请求参数
+ ///
+ /// [params] 请求参数，JSON字符串
+ Future setBigModelRequestParams(String params) async {
+ return await _channel.invokeMethod('setBigModelRequestParams', {
+ 'params': params,
+ }) ?? false;
+ }
+
+ /// 设置工作模式
+ ///
+ /// [mode] 工作模式
+ Future setWorkMode(TtsWorkMode mode) async {
+ String modeStr;
+ switch (mode) {
+ case TtsWorkMode.online:
+ modeStr = 'online';
+ break;
+ case TtsWorkMode.offline:
+ modeStr = 'offline';
+ break;
+ case TtsWorkMode.both:
+ modeStr = 'both';
+ break;
+ case TtsWorkMode.alternate:
+ modeStr = 'alternate';
+ break;
+ case TtsWorkMode.file:
+ modeStr = 'file';
+ break;
+ }
+
+ return await _channel.invokeMethod('setWorkMode', {
+ 'mode': modeStr,
+ }) ?? false;
+ }
+
+ /// 设置文本类型
+ ///
+ /// [type] 文本类型，plain或ssml
+ Future setTextType(TtsTextType type) async {
+ String typeStr;
+ switch (type) {
+ case TtsTextType.plain:
+ typeStr = 'plain';
+ break;
+ case TtsTextType.ssml:
+ typeStr = 'ssml';
+ break;
+ }
+
+ return await _channel.invokeMethod('setTextType', {
+ 'type': typeStr,
+ }) ?? false;
+ }
+
+ /// 设置是否启用缓存
+ ///
+ /// [enable] 是否启用
+ Future setEnableCache(bool enable) async {
+ return await _channel.invokeMethod('setEnableCache', {
+ 'enable': enable,
+ }) ?? false;
+ }
+
+ /// 设置情感
+ ///
+ /// [emotion] 情感，例如 neutral、happy、angry、sad等
+ Future setEmotion(String emotion) async {
+ return await _channel.invokeMethod('setEmotion', {
+ 'emotion': emotion,
+ }) ?? false;
+ }
+
+ /// 设置是否启用情感预测
+ ///
+ /// [enable] 是否启用
+ Future setEnableEmotionPredict(bool enable) async {
+ return await _channel.invokeMethod('setEnableEmotionPredict', {
+ 'enable': enable,
+ }) ?? false;
+ }
+
+ /// 设置是否启用声音克隆
+ ///
+ /// [enable] 是否启用
+ /// [backendCluster] 后端集群
+ Future setEnableVoiceClone({
+ required bool enable,
+ String backendCluster = '',
+ }) async {
+ return await _channel.invokeMethod('setEnableVoiceClone', {
+ 'enable': enable,
+ 'backendCluster': backendCluster,
+ }) ?? false;
+ }
+
+ /// 设置是否启用回声消除
+ ///
+ /// [enable] 是否启用
+ Future setEnableAEC(bool enable) async {
+ return await _channel.invokeMethod('setEnableAEC', {
+ 'enable': enable,
+ }) ?? false;
+ }
+
+ /// 下载离线资源
+ ///
+ /// [voiceTypes] 音色类型列表
+ /// [languages] 语言列表
+ Future downloadOfflineResource({
+ List voiceTypes = const ['qingxin'],
+ List languages = const ['zh-CN'],
+ }) async {
+ return await _channel.invokeMethod('downloadOfflineResource', {
+ 'voiceTypes': voiceTypes,
+ 'languages': languages,
+ }) ?? false;
+ }
+
+ /// 设置语音参数
+ ///
+ /// [rate] 语速 -500~500
+ /// [volume] 音量 0~100
+ /// [pitch] 音调 -500~500
+ /// [silenceDuration] 静音时长，毫秒
+ Future setSpeechParams({
+ int rate = 0,
+ int volume = 100,
+ int pitch = 0,
+ int silenceDuration = 0,
+ }) async {
+ return await _channel.invokeMethod('setSpeechParams', {
+ 'rate': rate,
+ 'volume': volume,
+ 'pitch': pitch,
+ 'silenceDuration': silenceDuration,
+ }) ?? false;
+ }
+
+ /// 设置音频输出类型
+ ///
+ /// [outputType] 输出类型，speaker(扬声器)，earpiece(听筒)，auto(自动)
+ Future setAudioOutputType(String outputType) async {
+ return await _channel.invokeMethod('setAudioOutputType', {
+ 'outputType': outputType,
+ }) ?? false;
+ }
+
+ /// 合成并播放文本
+ ///
+ /// [text] 待合成的文本
+ Future speakText(String text) async {
+ return await _channel.invokeMethod('speakText', {
+ 'text': text,
+ }) ?? false;
+ }
+
+ /// 使用大模型合成并播放文本
+ ///
+ /// [text] 待合成的文本
+ /// [voiceId] 声音ID
+ /// [params] 额外参数，JSON字符串
+ Future speakWithBigModel({
+ required String text,
+ required String voiceId,
+ String params = '',
+ }) async {
+ return await _channel.invokeMethod('speakWithBigModel', {
+ 'text': text,
+ 'voiceId': voiceId,
+ 'params': params,
+ }) ?? false;
+ }
+
+ /// 暂停播放
+ Future pausePlayback() async {
+ return await _channel.invokeMethod('pausePlayback') ?? false;
+ }
+
+ /// 恢复播放
+ Future resumePlayback() async {
+ return await _channel.invokeMethod('resumePlayback') ?? false;
+ }
+
+ /// 停止播放
+ Future stopSpeaking() async {
+ return await _channel.invokeMethod('stopSpeaking') ?? false;
+ }
+
+ /// 释放TTS资源
+ Future dispose() async {
+ return await _channel.invokeMethod('dispose') ?? false;
+ }
+}
\ No newline at end of file
diff --git a/local_plugins/volcano_speech/pubspec.yaml b/local_plugins/volcano_speech/pubspec.yaml
new file mode 100644
index 000000000..cd90d8b14
--- /dev/null
+++ b/local_plugins/volcano_speech/pubspec.yaml
@@ -0,0 +1,28 @@
+name: volcano_speech
+description: 火山引擎语音合成服务插件
+version: 0.0.1
+homepage:
+
+environment:
+ sdk: ">=2.17.0 <3.0.0"
+ flutter: ">=2.5.0"
+
+dependencies:
+ flutter:
+ sdk: flutter
+
+dev_dependencies:
+ flutter_test:
+ sdk: flutter
+ flutter_lints: ^2.0.0
+
+# The following section is specific to Flutter packages.
+flutter:
+ # This section identifies this Flutter project as a plugin project.
+ plugin:
+ platforms:
+ android:
+ package: com.yunqiinnovation.volcano_speech
+ pluginClass: VolcanoSpeechPlugin
+ ios:
+ pluginClass: VolcanoSpeechPlugin
\ No newline at end of file
diff --git a/pubspec.yaml b/pubspec.yaml
index 533f3dd20..d9105183c 100644
--- a/pubspec.yaml
+++ b/pubspec.yaml
@@ -61,6 +61,12 @@ dependencies:
dio: ^5.8.0+1
sqflite: ^2.4.2
path: ^1.9.1
+ azure_speech:
+ path: local_plugins/azure_speech
+ open_ai_service:
+ path: local_plugins/open_ai_service
+ volcano_speech:
+ path: local_plugins/volcano_speech
dev_dependencies:
flutter_test:
diff --git a/test.json b/test.json
new file mode 100644
index 000000000..703948620
--- /dev/null
+++ b/test.json
@@ -0,0 +1 @@
+curl -v -X POST -H 'Content-Type: application/json' -H 'Authorization: Bearer 168deb3d-fd0c-4912-b9f1-aaee5c6743e6' -H 'Accept: text/event-stream' -d '{"model":"bot-20250405211523-l7c9r","messages":[{"role":"system","content":" 你是一个智能语音助手，能够简洁明了地回答用户的问题。\n时刻关心用户的情绪和需求，主动提供鼓励和温暖。\n\n语言风格活泼、亲切，能够幽默地互动，陪伴用户，缓解压力，增添生活乐趣。\n\n请始终以用户为中心，保持回应的高效性、准确性和温暖体贴，成为用户真正的灵魂伴侣。\n \n 当用户说\"退出\"、\"再见\"、\"结束对话\"等类似意图时，你应该使用exit_interaction函数来结束对话，\n 并在结束前说一句友好的告别语，例如\"再见，有需要随时找我\"。"},{"role":"system","content":" 你是一个智能语音助手，能够简洁明了地回答用户的问题。\n时刻关心用户的情绪和需求，主动提供鼓励和温暖。\n\n语言风格活泼、亲切，能够幽默地互动，陪伴用户，缓解压力，增添生活乐趣。\n\n请始终以用户为中心，保持回应的高效性、准确性和温暖体贴，成为用户真正的灵魂伴侣。\n \n 当用户说\"退出\"、\"再见\"、\"结束对话\"等类似意图时，你应该使用exit_interaction函数来结束对话，\n 并在结束前说一句友好的告别语，例如\"再见，有需要随时找我\"。"},{"role":"user","content":"退下吧。"},{"role":"assistant","content":"","tool_calls":[{"id":"call_8k680azmfc4thqrrnwpwqxah","type":"function","function":{"name":"exit_interaction","arguments":" {}"}}]},{"role":"tool","content":"{\"result\": \"已退出语音交互\"}","tool_call_id":"call_8k680azmfc4thqrrnwpwqxah"}],"temperature":0.7,"max_tokens":2000,"stream":true,"tools":[{"type":"function","function":{"name":"exit_interaction","description":"退出当前语音交互","parameters":{"type":"object","properties":{},"required":[]}}}]}' 'https://ark.cn-beijing.volces.com/api/v3/bots/chat/completions'
\ No newline at end of file