上传tts的优化方案

lib/data/services/speech_impl/azure_tts_service.dart

@@ -37,7 +37,7 @@ class AzureTtsService extends GetxService implements TtsService {
   // 可观察状态
   final isEnabled = true.obs;
   final _isSpeaking = false.obs;
-
+  String _currsessionid = '';
   // 流式文本缓冲区
   String _streamBuffer = '';
 
@@ -183,13 +183,32 @@ class AzureTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> speakOnce(String text) async {
+  Future<bool> startspeak(String sessionid) async {
+    if (!_isInitialized) await initialize();
+    if (!isEnabled.value) return false;
+
+    try {
+      // 开始合成，传递参数
+      final result = await _channel.invokeMethod('startspeak', {
+        'sessionid': sessionid,
+      });
+
+      return result == true;
+    } catch (e) {
+      Logger.error('语音合成失败: ${e.toString()}');
+      return false;
+    }
+  }
+
+  @override
+  Future<bool> speakOnce(String sessionid, String text) async {
     if (!_isInitialized) await initialize();
     if (!isEnabled.value || text.isEmpty) return false;
 
     try {
       // 开始合成，传递参数
       final result = await _channel.invokeMethod('speakText', {
+        'sessionid': sessionid,
         'text': text,
       });
 
@@ -201,7 +220,7 @@ class AzureTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> speakStream(String text) async {
+  Future<bool> speakStream(String sessionid, String text) async {
     if (!isEnabled.value || text.isEmpty) return false;
 
     try {
@@ -256,12 +275,12 @@ class AzureTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> flushStream() async {
+  Future<bool> flushStream(String sessionid) async {
     if (_textQueue.isEmpty && _streamBuffer.isEmpty) return true;
 
     try {
       // 处理缓冲区中可能的完整句子
-      await speakStream('');
+      await speakStream(sessionid, '');
 
       // 如果缓冲区仍有剩余文本，将其添加到播放队列
       if (_streamBuffer.isNotEmpty && _streamBuffer.trim().isNotEmpty) {
@@ -345,7 +364,7 @@ class AzureTtsService extends GetxService implements TtsService {
       final text = _textQueue.first;
 
       // 开始合成
-      final success = await speakOnce(text);
+      final success = await speakOnce(_currsessionid, text);
 
       // 仅当合成操作失败时才从队列移除并继续处理下一项
       // 成功的话会等待合成完成事件后自动处理下一项

lib/data/services/speech_impl/voice_clone_tts_service.dart

@@ -27,6 +27,7 @@ class VoiceCloneTtsService extends GetxService implements TtsService {
   late ConcatenatingAudioSource _playlist;
 
   // 当前语音
+  String _currsessionid = '';
   final RxString _currentVoice = ''.obs;
 
   @override
@@ -215,7 +216,21 @@ class VoiceCloneTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> speakOnce(String text) async {
+  Future<bool> startspeak(String sessionid) async {
+    if (!_isInitialized) await initialize();
+    if (!isEnabled.value) return false;
+
+    try {
+      _currsessionid = sessionid;
+      return true;
+    } catch (e) {
+      Logger.error('语音合成失败: ${e.toString()}');
+      return false;
+    }
+  }
+
+  @override
+  Future<bool> speakOnce(String sessionid, String text) async {
     if (!isEnabled.value || text.isEmpty) return false;
 
     try {
@@ -236,7 +251,7 @@ class VoiceCloneTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> speakStream(String text) async {
+  Future<bool> speakStream(String sessionid, String text) async {
     if (!isEnabled.value || text.isEmpty) return false;
 
     try {
@@ -295,12 +310,14 @@ class VoiceCloneTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> flushStream() async {
+  Future<bool> flushStream(
+    String sessionid,
+  ) async {
     if (_textQueue.isEmpty && _streamBuffer.isEmpty) return true;
 
     try {
       // 处理缓冲区中可能的完整句子
-      await speakStream('');
+      await speakStream(sessionid, '');
 
       // 如果缓冲区仍有剩余文本，将其添加到播放队列
       if (_streamBuffer.isNotEmpty && _streamBuffer.trim().isNotEmpty) {
@@ -344,7 +361,7 @@ class VoiceCloneTtsService extends GetxService implements TtsService {
         final item = _textQueue.removeAt(0);
 
         // 使用speakOnce播放当前项
-        await speakOnce(item);
+        await speakOnce(_currsessionid, item);
       }
     } catch (e) {
       Logger.error('处理语音队列出错: ${e.toString()}');

lib/data/services/speech_impl/volcano_tts_service.dart

@@ -206,7 +206,26 @@ class VolcanoTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> speakOnce(String text) async {
+  Future<bool> startspeak(String sessionid) async {
+    if (!await _ensureInitialized()) return false;
+
+    try {
+      // 单次播放
+      final result = await _channel.invokeMethod<bool>('startspeak', {
+            'sessionid': sessionid,
+          }) ??
+          false;
+
+      return result;
+    } catch (e) {
+      debugPrint('单次播放失败：$e');
+      _eventStreamController.add(TtsEvent.error('单次播放失败：$e'));
+      return false;
+    }
+  }
+
+  @override
+  Future<bool> speakOnce(String sessionid, String text) async {
     if (!await _ensureInitialized()) return false;
 
     try {
@@ -225,7 +244,7 @@ class VolcanoTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> speakStream(String text) async {
+  Future<bool> speakStream(String sessionid, String text) async {
     if (!await _ensureInitialized()) return false;
 
     try {
@@ -244,7 +263,9 @@ class VolcanoTtsService extends GetxService implements TtsService {
   }
 
   @override
-  Future<bool> flushStream() async {
+  Future<bool> flushStream(
+    String sessionid,
+  ) async {
     if (!await _ensureInitialized()) return false;
 
     try {

lib/data/services/tts_service.dart

@@ -19,14 +19,17 @@ abstract class TtsService {
   /// 设置语音
   Future<bool> setVoice(String voiceName);
 
+  ///开始新的播放会话
+  Future<bool> startspeak(String sessionid);
+
   /// 单次播放文本，完成后返回
-  Future<bool> speakOnce(String text);
+  Future<bool> speakOnce(String sessionid, String text);
 
   /// 将文本加入播放队列，支持连续播放多段文本
-  Future<bool> speakStream(String text);
+  Future<bool> speakStream(String sessionid, String text);
 
   /// 播放队列中剩余文本
-  Future<bool> flushStream();
+  Future<bool> flushStream(String sessionid);
 
   /// 停止播放
   Future<bool> stop();

lib/modules/FTFTranslation/controllers/FTFTranslation_controller.dart

@@ -42,7 +42,7 @@ class FTFTranslationController extends GetxController
   final sourceLanguageCode = 'zh-CN'.obs;
   final targetLanguageCode = 'en-US'.obs;
   final isTtsEnabled = true.obs;
-
+  var _currsessionid = '';
   @override
   void onInit() {
     super.onInit();
@@ -168,7 +168,7 @@ class FTFTranslationController extends GetxController
           'en-US-AriaNeural';
 
       await _ttsService.setVoice(voiceName);
-      await _ttsService.speakOnce(text);
+      await _ttsService.speakOnce(_currsessionid, text);
     } catch (e) {
       Logger.error('播放翻译失败: ${e.toString()}');
     }

lib/modules/agent/controllers/agent_controller.dart

@@ -949,10 +949,6 @@ class AgentController extends GetxController {
     final text = textController.text.trim();
     if (text.isEmpty) return;
 
-    final message = Message(isUser: true, text: text);
-    messages.add(message);
-    textController.clear();
-
     // 使用reverse:true时通常不需要手动滚动
     // 3. 调用AI服务处理
     try {
@@ -963,6 +959,9 @@ class AgentController extends GetxController {
       // 根据TTS状态决定是否朗读
       await AgentService.processTextInput(text,
           speakResponse: isTtsEnabled.value);
+      final message = Message(isUser: true, text: text);
+      messages.add(message);
+      textController.clear();
     } catch (e) {
       Logger.e(TAG, '发送文本消息失败: $e');
       isProcessing.value = false;

lib/modules/speech_test/controllers/speech_test_controller.dart

@@ -53,6 +53,8 @@ class SpeechTestController extends GetxController {
     '这是第三句测试语音，将在连续模式下播放完毕。'
   ];
 
+  var _currsessionid = '';
+
   @override
   void onInit() {
     super.onInit();
@@ -246,7 +248,7 @@ class SpeechTestController extends GetxController {
       isProcessing.value = true;
       ttsStatus.value = '准备播放...';
 
-      final result = await _ttsService.speakOnce(text);
+      final result = await _ttsService.speakOnce(_currsessionid, text);
 
       if (!result) {
         ttsStatus.value = '播放失败';
@@ -269,7 +271,7 @@ class SpeechTestController extends GetxController {
 
       // 为避免阻塞，使用一个句子接一个句子的方式添加到队列
       for (final sentence in testSentences) {
-        if (!await _ttsService.speakStream(sentence)) {
+        if (!await _ttsService.speakStream(_currsessionid, sentence)) {
           ttsStatus.value = '添加句子到队列失败';
           break;
         }
@@ -278,7 +280,7 @@ class SpeechTestController extends GetxController {
       }
 
       // 刷新队列，确保播放完成
-      await _ttsService.flushStream();
+      await _ttsService.flushStream(_currsessionid);
     } catch (e) {
       ttsStatus.value = '连续播放异常: $e';
       Logger.error('连续播放异常: ${e.toString()}');

lib/modules/translation/controllers/translation_controller.dart

@@ -88,6 +88,7 @@ class TranslationController extends GetxController {
   final targetLanguageCode = 'en-US'.obs;
   var detectedLanguageCode = '';
 
+  var currsessionid = '';
   final currentModeTitle = 'simultaneousTranslation'.obs;
   final currentMode = 'simultaneous'.obs;
   // 翻译历史和当前项
@@ -935,7 +936,7 @@ class TranslationController extends GetxController {
 
       final startTime = DateTime.now();
       await _ttsService.setVoice(voiceName);
-      await _ttsService.speakOnce(text);
+      await _ttsService.speakOnce(currsessionid, text);
       final endTime = DateTime.now();
 
       // 记录TTS API调用统计 - 优化版（基于文本字符数计费）
@@ -991,7 +992,7 @@ class TranslationController extends GetxController {
           _languageManager.getTtsVoiceNameByAsrCode(item.targetLanguageCode) ??
               'en-US-AriaNeural';
       await _ttsService.setVoice(voiceName);
-      await _ttsService.speakOnce(item.translatedText);
+      await _ttsService.speakOnce(currsessionid, item.translatedText);
     } catch (e) {
       Logger.error('播放翻译项失败: ${e.toString()}');
     }
@@ -1010,7 +1011,7 @@ class TranslationController extends GetxController {
               'en-US-AriaNeural';
 
       await _ttsService.setVoice(voiceName);
-      await _ttsService.speakOnce(text);
+      await _ttsService.speakOnce(currsessionid, text);
     } catch (e) {
       Logger.error('播放文本失败: ${e.toString()}');
     }

local_plugins/agent_service/ios/agent_service/Sources/agent_service/AgentServiceImpl.swift

@@ -378,7 +378,7 @@ class AgentServiceImpl: NSObject {
                 return false
             }
         }
-        
+        stopTts();
         if isRecognizing {
             stopRecognition()
         }
@@ -456,7 +456,7 @@ audioStream.saveAudioDataTo(data: audioData)
         return true
     }
     
-    func speakText(_ text: String) -> Bool {
+    func speakText(sessionid sessionid:String,_ text: String) -> Bool {
         if !isInitialized {
             sendError("服务未初始化", code: "NOT_INITIALIZED")
             return false
@@ -466,7 +466,7 @@ audioStream.saveAudioDataTo(data: audioData)
             return false
         }
         
-        return azureTtsHelper?.speakOnce(text) ?? false
+        return azureTtsHelper?.speakOnce(sessionid:sessionid,text) ?? false
     }
     
     func stopTts() -> Bool {
@@ -591,6 +591,7 @@ audioStream.saveAudioDataTo(data: audioData)
             imagePath: imagePath
         )
         currsessionId = UUID().uuidString
+        azureTtsHelper?.startspeak(sessionid: currsessionId)
         os_log("设置回调并调用sendMessageStream", log: logger, type: .info)
         chatApiService.setStreamCallback(currsessionId,callback)
         chatApiService.sendMessageStream(currsessionId,messages: messages)
@@ -1253,7 +1254,7 @@ class ChatApiStreamCallback: StreamCallback {
             responseBuilder += token
             
             if speakResponse && reply && broadcast{
-                try agentService.azureTtsHelper?.speakStream(token)
+                try agentService.azureTtsHelper?.speakStream(sessionid:sessionId,token)
                 // 在开始流式TTS时立即停止气泡音
             }
             if (reply && token != "") {
@@ -1275,7 +1276,7 @@ class ChatApiStreamCallback: StreamCallback {
           
             
             if speakResponse && reply && broadcast && sessionId == agentService.currsessionId{
-                agentService.azureTtsHelper?.flushStream()
+                agentService.azureTtsHelper?.flushStream(sessionid:sessionId)
             }
             
             let response = responseBuilder

local_plugins/agent_service/ios/agent_service/Sources/agent_service/AgentServicePlugin.swift

@@ -86,12 +86,13 @@ public class AgentServicePlugin: NSObject, FlutterPlugin {
             
         case "speakText":
             guard let arguments = call.arguments as? [String: Any],
+                  let sessionid = arguments["sessionid"] as? String,
                   let text = arguments["text"] as? String else {
                 result(FlutterError(code: "INVALID_ARGUMENTS", message: "文本不能为空", details: nil))
                 return
             }
             
-            let success = impl.speakText(text)
+            let success = impl.speakText(sessionid:sessionid,text)
             result(success)
             
         case "stopTts":

local_plugins/azure_speech/ios/azure_speech/Sources/azure_speech/AzureSpeechPlugin.swift

@@ -480,26 +480,33 @@ private func sendAstEvent(_ event: [String: Any]) {
             
         case "speakText":
             guard let args = call.arguments as? [String: Any],
+                  let sessionid = args["sessionid"] as? String,
                   let text = args["text"] as? String else {
                 result(FlutterError(code: "INVALID_ARGUMENTS", message: "文本不能为空", details: nil))
                 return
             }
             
-            let success = azureTtsHelper.speakOnce(text)
+            let success = azureTtsHelper.speakOnce(sessionid:sessionid,text)
             result(success)
             
         case "speakStream":
             guard let args = call.arguments as? [String: Any],
+                  let sessionid = args["sessionid"] as? String,
                   let text = args["text"] as? String else {
                 result(FlutterError(code: "INVALID_ARGUMENTS", message: "文本不能为空", details: nil))
                 return
             }
             
-            let success = azureTtsHelper.speakStream(text)
+            let success = azureTtsHelper.speakStream(sessionid:sessionid,text)
             result(success)
             
         case "flushStream":
-            let success = azureTtsHelper.flushStream()
+            guard let args = call.arguments as? [String: Any],
+                  let sessionid = args["sessionid"] as? String else {
+                result(FlutterError(code: "INVALID_ARGUMENTS", message: "文本不能为空", details: nil))
+                return
+            }
+            let success = azureTtsHelper.flushStream(sessionid: sessionid)
             result(success)
             
         case "stopSpeaking":

local_plugins/azure_speech/ios/azure_speech/Sources/azure_speech/AzureTtsHelper.swift

@@ -16,6 +16,7 @@ public class AzureTtsHelper: NSObject, ITtsService {
     private var synthesizer: SPXSpeechSynthesizer?
     private var isInitialized = false
     private var speaking = false
+    private var sessionid = ""  //语音播报任务id
     
     // 当前配置
     private var currentVoice = DEFAULT_VOICE
@@ -109,10 +110,14 @@ public class AzureTtsHelper: NSObject, ITtsService {
         }
     }
     
+    public func startspeak(sessionid sessionid: String)-> Bool {
+        self.sessionid = sessionid
+        return true
+    }
     /**
      * 单次合成并播放
      */
-    public func speakOnce(_ text: String) -> Bool {
+    public func speakOnce(sessionid sessionid: String,_ text: String) -> Bool {
         if !isInitialized {
             os_log("语音合成未初始化", log: log, type: .error)
             notifyEvent(eventType: .error, params: [
@@ -122,6 +127,10 @@ public class AzureTtsHelper: NSObject, ITtsService {
             return false
         }
         
+        if (self.sessionid != sessionid){
+            return false
+        }
+        
         // 清理文本
         let cleanedText = cleanTextForTTS(text)
         if cleanedText.isEmpty {
@@ -130,7 +139,7 @@ public class AzureTtsHelper: NSObject, ITtsService {
         
         // 异步处理
         enqueueSynthesisTask {
-            self.performSynthesis(text: cleanedText)
+            self.performSynthesis(sessionid: sessionid, text: cleanedText)
         }
         
         return true
@@ -139,7 +148,7 @@ public class AzureTtsHelper: NSObject, ITtsService {
     /**
      * 流式合成文本
      */
-    public func speakStream(_ text: String) -> Bool {
+    public func speakStream(sessionid sessionid: String,_ text: String) -> Bool {
         if !isInitialized {
             notifyEvent(eventType: .error, params: [
                 "errorCode": "NOT_INITIALIZED",
@@ -147,6 +156,10 @@ public class AzureTtsHelper: NSObject, ITtsService {
             ])
             return false
         }
+        if (self.sessionid != sessionid){
+            return false
+        }
+        self.sessionid = sessionid
         
         if text.isEmpty {
             return true
@@ -184,7 +197,7 @@ public class AzureTtsHelper: NSObject, ITtsService {
             streamBuffer = String(currentText[startIndex...])
             
             if !textToSpeak.isEmpty {
-                return speakOnce(textToSpeak)
+                return speakOnce(sessionid: sessionid, textToSpeak)
             }
         }
         
@@ -194,7 +207,7 @@ public class AzureTtsHelper: NSObject, ITtsService {
     /**
      * 刷新并播放流式文本
      */
-    public func flushStream() -> Bool {
+    public func flushStream(sessionid sessionid: String) -> Bool {
         if !isInitialized {
             notifyEvent(eventType: .error, params: [
                 "errorCode": "NOT_INITIALIZED",
@@ -210,7 +223,7 @@ public class AzureTtsHelper: NSObject, ITtsService {
             return true
         }
         
-        return speakOnce(remainingText)
+        return speakOnce(sessionid:sessionid,remainingText)
     }
     
     /**
@@ -218,14 +231,15 @@ public class AzureTtsHelper: NSObject, ITtsService {
      */
     public func stop() -> Bool {
         speaking = false
+        sessionid = ""
         streamBuffer = ""
-        
+        lastSpokenText = ""
         taskLock.lock()
         pendingTasks.removeAll()
         taskLock.unlock()
         
-        synthesisQueue.async {
         do {
+            print("liwei-------------停止语音合成播放")
             try self.synthesizer?.stopSpeaking()
             self.notifyEvent(eventType: .synthesisCanceled)
         } catch {
@@ -235,6 +249,14 @@ public class AzureTtsHelper: NSObject, ITtsService {
                 "errorMessage": error.localizedDescription
             ])
         }
+        
+        synthesisQueue.async {
+            do {
+                self.synthesizer = nil
+                try self.recreateSynthesizer()
+            }catch {
+                os_log("重置语音合成失败: %{public}@", log: self.log, type: .error, error.localizedDescription)
+            }
         }
         
         return true
@@ -314,7 +336,7 @@ public class AzureTtsHelper: NSObject, ITtsService {
             
         // 如果之前正在播放，恢复播放
         if wasSpeaking, let lastText = lastSpokenText {
-            speakOnce(lastText)
+            speakOnce(sessionid: "",lastText)
         }
         
         return true
@@ -386,13 +408,19 @@ public class AzureTtsHelper: NSObject, ITtsService {
     /**
      * 执行语音合成
      */
-    private func performSynthesis(text: String) {
+    private func performSynthesis(sessionid:String,text: String) {
+        if (self.sessionid != sessionid) {
+            return
+        }
         speaking = true
-        lastSpokenText = text
+//        lastSpokenText = text
         // 生成SSML
         let ssml = generateOptimizedSsml(text)
         
         do {
+            if (self.sessionid != sessionid) {
+                return
+            }
             os_log("开始合成: %{public}@", log: log, type: .debug, ssml)
             let result = try synthesizer?.startSpeakingSsml(ssml)
             

local_plugins/speech/ios/speech/Sources/speech/ITtsService.swift

@@ -92,7 +92,11 @@ import Foundation
      * @param text 要合成的文本
      * @return 是否成功开始合成
      */
-    func speakOnce(_ text: String) -> Bool
+    func speakOnce(sessionid sessionid: String,_ text: String) -> Bool
+    
+    
+    //开始播报
+    func startspeak(sessionid sessionid: String)-> Bool
     
     /**
      * 流式合成
@@ -101,14 +105,14 @@ import Foundation
      * @param text 要合成的文本片段
      * @return 是否成功处理
      */
-    func speakStream(_ text: String) -> Bool
+    func speakStream(sessionid sessionid: String,_ text: String) -> Bool
     
     /**
      * 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
+    func flushStream(sessionid sessionid: String) -> Bool
     
     /**
      * 停止当前合成和播放