You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
497 lines
16 KiB
497 lines
16 KiB
import Foundation
|
|
import AVFoundation
|
|
import MicrosoftCognitiveServicesSpeech
|
|
// 自定义语音处理组件，提供音频流处理等功能
|
|
import speech
|
|
|
|
/**
|
|
* Azure TTS Helper
|
|
*
|
|
* 基于微软Azure语音服务的TTS实现
|
|
* 参考文档: https://learn.microsoft.com/en-us/azure/ai-services/speech-service/how-to-speech-synthesis
|
|
*/
|
|
class AzureTtsHelper: NSObject, ITtsService {
|
|
private let tag = "AzureTtsHelper"
|
|
|
|
private static let DEFAULT_LANGUAGE = "zh-CN"
|
|
private static let DEFAULT_VOICE = "zh-CN-XiaoxiaoNeural"
|
|
|
|
// Azure语音服务配置
|
|
private var speechConfig: SPXSpeechConfiguration?
|
|
private var synthesizer: SPXSpeechSynthesizer?
|
|
private var isInitialized = false
|
|
private var speaking = false
|
|
|
|
// 当前配置
|
|
private var currentVoice = DEFAULT_VOICE
|
|
private var currentLanguage = DEFAULT_LANGUAGE
|
|
private var currentRate = "+0%"
|
|
private var currentPitch = "+0%"
|
|
private var currentVolume = "100%"
|
|
|
|
// 事件监听器列表
|
|
private var eventListeners = NSHashTable<AnyObject>.weakObjects()
|
|
|
|
// 流式文本处理的缓冲区
|
|
private var streamBuffer = ""
|
|
private var lastSpeakTime: TimeInterval = 0
|
|
|
|
// 音频会话
|
|
private let audioSession = AVAudioSession.sharedInstance()
|
|
|
|
/**
|
|
* 初始化语音合成服务
|
|
*
|
|
* @param appId 服务应用ID
|
|
* @param token 服务访问令牌/订阅密钥
|
|
* @param resource 服务资源ID/区域（可选）
|
|
* @param language 语言代码，如"zh-CN"
|
|
* @return 是否初始化成功
|
|
*/
|
|
func initialize(ttsAppId: String, ttsAppToken: String, ttsResource: String, language: String) -> Bool {
|
|
do {
|
|
// 创建语音配置
|
|
if ttsResource.isEmpty {
|
|
speechConfig = try SPXSpeechConfiguration(subscription: ttsAppToken, region: "eastasia")
|
|
} else {
|
|
speechConfig = try SPXSpeechConfiguration(subscription: ttsAppToken, region: ttsResource)
|
|
}
|
|
|
|
// 设置语言
|
|
speechConfig?.speechSynthesisLanguage = language
|
|
currentLanguage = language
|
|
|
|
// 设置音频输出格式 - 使用16k、16位的PCM格式
|
|
speechConfig?.setPropertyTo("Audio16Khz16BitMonoPcm",
|
|
byName: "SpeechServiceConnection_SynthOutputFormat")
|
|
|
|
// 创建合成器，使用默认音频输出（扬声器）
|
|
synthesizer = try SPXSpeechSynthesizer(speechConfig!)
|
|
|
|
// 设置事件监听
|
|
setupEventListeners()
|
|
|
|
// 设置默认音色 - 中文默认使用晓晓，英文默认使用Jenny
|
|
if language.lowercased().starts(with: "zh") {
|
|
_ = setVoice("zh-CN-XiaoxiaoNeural")
|
|
} else {
|
|
_ = setVoice("en-US-JennyNeural")
|
|
}
|
|
|
|
// 设置初始化完成
|
|
isInitialized = true
|
|
|
|
return true
|
|
} catch {
|
|
print("\(tag): 语音合成服务初始化失败: \(error.localizedDescription)")
|
|
return false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 设置语音角色
|
|
*
|
|
* @param voiceName 语音角色名称（不同服务的语音角色命名可能不同）
|
|
* @return 是否设置成功
|
|
*/
|
|
func setVoice(_ voiceName: String) -> Bool {
|
|
if !isInitialized { return false }
|
|
|
|
do {
|
|
currentVoice = voiceName
|
|
|
|
// 更新语音名称
|
|
if let config = speechConfig {
|
|
config.speechSynthesisVoiceName = voiceName
|
|
}
|
|
|
|
// 重新创建合成器
|
|
recreateSynthesizer()
|
|
return true
|
|
} catch {
|
|
print("\(tag): 设置语音失败: \(error.localizedDescription)")
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "VOICE_SET_FAILED",
|
|
"errorMessage": "设置语音失败: \(error.localizedDescription)"
|
|
])
|
|
return false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 单次合成并播放
|
|
*
|
|
* @param text 要合成的文本
|
|
* @return 是否成功开始合成
|
|
*/
|
|
func speakOnce(_ text: String) -> Bool {
|
|
if !isInitialized {
|
|
print("\(tag): 语音合成未初始化")
|
|
return false
|
|
}
|
|
|
|
// 已经在播放，先停止
|
|
if speaking {
|
|
_ = stop()
|
|
}
|
|
|
|
// 重置状态
|
|
speaking = true
|
|
|
|
// 生成SSML
|
|
let ssml = generateSsml(text)
|
|
|
|
do {
|
|
// 创建合成任务
|
|
try synthesizer?.startSpeakingSsml(ssml)
|
|
return true
|
|
} catch {
|
|
print("\(tag): 语音合成失败: \(error.localizedDescription)")
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "SYNTHESIS_FAILED",
|
|
"errorMessage": error.localizedDescription
|
|
])
|
|
speaking = false
|
|
return false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 流式合成文本
|
|
*
|
|
* @param text 要合成的文本片段
|
|
* @return 是否成功处理
|
|
*/
|
|
func speakStream(_ text: String) -> Bool {
|
|
if !isInitialized || text.isEmpty {
|
|
if !isInitialized {
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "NOT_INITIALIZED",
|
|
"errorMessage": "TTS引擎未初始化"
|
|
])
|
|
}
|
|
return false
|
|
}
|
|
|
|
do {
|
|
// 添加新文本到缓冲区
|
|
streamBuffer.append(text)
|
|
|
|
// 增加500ms防抖逻辑
|
|
let currentTime = Date().timeIntervalSince1970
|
|
if currentTime - lastSpeakTime < 0.6 {
|
|
return true
|
|
}
|
|
lastSpeakTime = currentTime
|
|
|
|
let currentText = streamBuffer
|
|
|
|
// 定义标点符号列表
|
|
let punctuationMarks: [Character] = [".", "。", "!", "！", "?", "？", ";", "；", ",", "，", ":", "：", "\n"]
|
|
|
|
// 查找最后一个标点符号的位置
|
|
var lastPunctuationIndex = -1
|
|
for (i, char) in currentText.enumerated().reversed() {
|
|
if punctuationMarks.contains(char) {
|
|
lastPunctuationIndex = i
|
|
break
|
|
}
|
|
}
|
|
|
|
// 如果找到标点符号，则播放到该标点符号
|
|
if lastPunctuationIndex >= 0 {
|
|
// 提取要播放的文本（包含标点符号）
|
|
let textToSpeak = String(currentText.prefix(lastPunctuationIndex + 1))
|
|
|
|
// 剩余的文本保存在缓冲区中
|
|
let startIndex = currentText.index(currentText.startIndex, offsetBy: lastPunctuationIndex + 1)
|
|
streamBuffer = String(currentText[startIndex...])
|
|
|
|
// 播放提取的文本
|
|
return speakOnce(textToSpeak)
|
|
}
|
|
|
|
// 如果没有找到标点符号，则等待更多文本
|
|
return true
|
|
} catch {
|
|
print("\(tag): 流式语音合成失败: \(error.localizedDescription)")
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "STREAM_FAILED",
|
|
"errorMessage": "流式语音合成失败: \(error.localizedDescription)"
|
|
])
|
|
return false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 刷新并播放流式文本缓冲区中的剩余内容
|
|
*
|
|
* @return 是否成功处理
|
|
*/
|
|
func flushStream() -> Bool {
|
|
if !isInitialized {
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "NOT_INITIALIZED",
|
|
"errorMessage": "TTS引擎未初始化"
|
|
])
|
|
return false
|
|
}
|
|
|
|
do {
|
|
// 获取缓冲区中剩余的文本
|
|
let remainingText = streamBuffer
|
|
|
|
// 清空缓冲区
|
|
streamBuffer = ""
|
|
|
|
// 如果缓冲区为空，直接返回成功
|
|
if remainingText.isEmpty {
|
|
return true
|
|
}
|
|
|
|
// 播放剩余文本
|
|
return speakOnce(remainingText)
|
|
} catch {
|
|
print("\(tag): 刷新流式文本失败: \(error.localizedDescription)")
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "FLUSH_FAILED",
|
|
"errorMessage": "刷新流式文本失败: \(error.localizedDescription)"
|
|
])
|
|
return false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 停止语音合成和播放
|
|
*
|
|
* @return 是否成功停止
|
|
*/
|
|
func stop() -> Bool {
|
|
if !speaking || synthesizer == nil {
|
|
return true
|
|
}
|
|
|
|
do {
|
|
// 停止合成
|
|
try synthesizer?.stopSpeaking()
|
|
speaking = false
|
|
return true
|
|
} catch {
|
|
print("\(tag): 停止语音合成失败: \(error.localizedDescription)")
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "STOP_FAILED",
|
|
"errorMessage": error.localizedDescription
|
|
])
|
|
return false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 释放资源
|
|
* 在不再需要服务时调用，释放底层资源
|
|
*/
|
|
func dispose() {
|
|
do {
|
|
// 停止播放
|
|
_ = stop()
|
|
|
|
// 释放合成器
|
|
synthesizer = nil
|
|
|
|
// 释放配置
|
|
speechConfig = nil
|
|
|
|
// 清空流缓冲区
|
|
streamBuffer = ""
|
|
|
|
// 重置状态
|
|
isInitialized = false
|
|
speaking = false
|
|
} catch {
|
|
// 即使发生异常，也确保重置状态
|
|
isInitialized = false
|
|
speaking = false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 添加TTS事件监听器
|
|
*
|
|
* @param listener 事件监听器
|
|
*/
|
|
func addListener(_ listener: TtsEventListener) {
|
|
eventListeners.add(listener as AnyObject)
|
|
}
|
|
|
|
/**
|
|
* 移除TTS事件监听器
|
|
*
|
|
* @param listener 要移除的事件监听器
|
|
*/
|
|
func removeListener(_ listener: TtsEventListener) {
|
|
eventListeners.remove(listener as AnyObject)
|
|
}
|
|
|
|
/**
|
|
* 添加音频数据监听器
|
|
* 由于不再支持自定义音频流，此方法实际上不再有效
|
|
*
|
|
* @param listener 音频数据监听器
|
|
*/
|
|
func addAudioDataListener(_ listener: AudioDataListener) {
|
|
print("\(tag): 警告：不支持音频数据监听器功能")
|
|
}
|
|
|
|
/**
|
|
* 移除音频数据监听器
|
|
* 由于不再支持自定义音频流，此方法实际上不再有效
|
|
*
|
|
* @param listener 要移除的音频数据监听器
|
|
*/
|
|
func removeAudioDataListener(_ listener: AudioDataListener) {
|
|
// 不做任何操作
|
|
}
|
|
|
|
/**
|
|
* 当前是否正在播放/合成
|
|
*/
|
|
func isSpeaking() -> Bool {
|
|
return speaking
|
|
}
|
|
|
|
// MARK: - 辅助方法
|
|
|
|
/**
|
|
* 设置事件监听器
|
|
*/
|
|
private func setupEventListeners() {
|
|
guard let synthesizer = synthesizer else { return }
|
|
|
|
// 添加合成开始事件处理器
|
|
synthesizer.addSynthesisStartedEventHandler { [weak self] _, _ in
|
|
guard let self = self else { return }
|
|
self.notifyEvent(eventType: .synthesisStarted)
|
|
}
|
|
|
|
// 添加合成中事件处理器
|
|
synthesizer.addSynthesizingEventHandler { [weak self] _, _ in
|
|
// 可以在这里处理合成中的事件，目前没有特别操作
|
|
}
|
|
|
|
// 添加合成完成事件处理器
|
|
synthesizer.addSynthesisCompletedEventHandler { [weak self] _, e in
|
|
guard let self = self else { return }
|
|
self.speaking = false
|
|
self.notifyEvent(eventType: .synthesisCompleted)
|
|
}
|
|
|
|
// 添加合成取消事件处理器
|
|
synthesizer.addSynthesisCanceledEventHandler { [weak self] _, e in
|
|
guard let self = self else { return }
|
|
|
|
print("\(self.tag): 语音合成取消")
|
|
self.speaking = false
|
|
|
|
var params: [String: Any] = [:]
|
|
do {
|
|
let cancellationDetails = try SPXSpeechSynthesisCancellationDetails(fromCanceledSynthesisResult: e.result)
|
|
if cancellationDetails.reason == SPXCancellationReason.error {
|
|
params["reason"] = String(describing: cancellationDetails.reason.rawValue)
|
|
params["errorDetails"] = cancellationDetails.errorDetails ?? "未知错误"
|
|
}
|
|
} catch {
|
|
params["errorDetails"] = "获取取消详情失败: \(error.localizedDescription)"
|
|
}
|
|
|
|
self.notifyEvent(eventType: .synthesisCanceled, params: params)
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 触发事件通知
|
|
*/
|
|
private func notifyEvent(eventType: TtsEventType, params: [String: Any] = [:]) {
|
|
let event = TtsEvent(type: eventType, params: params)
|
|
|
|
// 在主线程通知事件
|
|
DispatchQueue.main.async {
|
|
for case let listener as TtsEventListener in self.eventListeners.allObjects {
|
|
listener.onEvent(event)
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 重新创建合成器
|
|
*/
|
|
private func recreateSynthesizer() {
|
|
do {
|
|
// 使用默认音频输出配置创建合成器
|
|
synthesizer = try SPXSpeechSynthesizer(speechConfig!)
|
|
|
|
// 设置事件监听
|
|
setupEventListeners()
|
|
} catch {
|
|
print("\(tag): 重新创建合成器失败: \(error.localizedDescription)")
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "RECREATE_FAILED",
|
|
"errorMessage": "重新创建合成器失败: \(error.localizedDescription)"
|
|
])
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 设置语音参数
|
|
*/
|
|
private func setSpeechParams(rate: Int = 0, pitch: Int = 0, volume: Int = 100) -> Bool {
|
|
if !isInitialized { return false }
|
|
|
|
do {
|
|
currentRate = formatPercentage(rate)
|
|
currentPitch = formatPercentage(pitch)
|
|
currentVolume = "\(min(max(volume, 0), 100))%"
|
|
return true
|
|
} catch {
|
|
print("\(tag): 设置语音参数失败: \(error.localizedDescription)")
|
|
notifyEvent(eventType: .error, params: [
|
|
"errorCode": "PARAMS_SET_FAILED",
|
|
"errorMessage": "设置语音参数失败: \(error.localizedDescription)"
|
|
])
|
|
return false
|
|
}
|
|
}
|
|
|
|
/**
|
|
* 格式化百分比值
|
|
*/
|
|
private func formatPercentage(_ value: Int) -> String {
|
|
return value >= 0 ? "+\(value)%" : "\(value)%"
|
|
}
|
|
|
|
/**
|
|
* 生成SSML
|
|
*/
|
|
private func generateSsml(_ rawText: String) -> String {
|
|
// 转义 XML 保留字符
|
|
var escapedText = rawText
|
|
.replacingOccurrences(of: "&", with: "&")
|
|
.replacingOccurrences(of: "<", with: "<")
|
|
.replacingOccurrences(of: ">", with: ">")
|
|
|
|
// 构造简化的SSML文档
|
|
let ssml = """
|
|
<speak version="1.0"
|
|
xmlns="http://www.w3.org/2001/10/synthesis"
|
|
xmlns:mstts="https://www.w3.org/2001/mstts"
|
|
xml:lang="zh-CN">
|
|
<voice name="\(currentVoice)">
|
|
<prosody rate="\(currentRate)" pitch="\(currentPitch)" volume="\(currentVolume)">
|
|
\(escapedText)
|
|
</prosody>
|
|
</voice>
|
|
</speak>
|
|
"""
|
|
|
|
return ssml
|
|
}
|
|
}
|
|
|