feat: 支持音色切换

backend/app/api/v1/endpoints/tts.py

@@ -340,55 +340,51 @@ async def save_audio_file(audio_data: bytes, filename: str) -> str:
 
 
 # 处理单个TTS任务
-async def process_tts_task(websocket, message_id: str, text: str):
+async def process_tts_task(websocket, message_id: str, text: str, speaker: str = None):
     """处理单个TTS任务（独立协程）"""
     tts_state = None
-    try:
-        print(f"开始处理TTS任务 [{message_id}]: {text}")
+    # 使用传入的speaker，如果没有则使用默认的
+    selected_speaker = speaker if speaker else SPEAKER
 
+    try:
+        print(f"开始处理TTS任务 [{message_id}]: {text}, 使用说话人: {selected_speaker}")
         # 获取TTS状态
         tts_state = tts_manager.get_tts_state(websocket, message_id)
         if not tts_state:
             raise Exception(f"找不到TTS状态: {message_id}")
-
         tts_state.is_processing = True
         # 生成音频文件名
         tts_state.audio_filename = generate_audio_filename()
-
         # 创建独立的TTS连接
         tts_state.volc_ws = await create_tts_connection()
-
         # 创建会话
         tts_state.session_id = uuid.uuid4().__str__().replace('-', '')
         tts_manager.register_session(tts_state.session_id, message_id)
-
         print(f"创建TTS会话 [{message_id}]: {tts_state.session_id}")
         header = Header(message_type=FULL_CLIENT_REQUEST,
                         message_type_specific_flags=MsgTypeFlagWithEvent,
                         serial_method=JSON).as_bytes()
         optional = Optional(event=EVENT_StartSession, sessionId=tts_state.session_id).as_bytes()
-        payload = get_payload_bytes(event=EVENT_StartSession, speaker=SPEAKER)
+        # 使用选择的speaker
+        payload = get_payload_bytes(event=EVENT_StartSession, speaker=selected_speaker)
         await send_event(tts_state.volc_ws, header, optional, payload)
-
         raw_data = await tts_state.volc_ws.recv()
         res = parser_response(raw_data)
         if res.optional.event != EVENT_SessionStarted:
             raise Exception("TTS会话启动失败")
         print(f"TTS会话创建成功 [{message_id}]: {tts_state.session_id}")
-
         # 发送文本到TTS服务
         print(f"发送文本到TTS服务 [{message_id}]...")
         header = Header(message_type=FULL_CLIENT_REQUEST,
                         message_type_specific_flags=MsgTypeFlagWithEvent,
                         serial_method=JSON).as_bytes()
         optional = Optional(event=EVENT_TaskRequest, sessionId=tts_state.session_id).as_bytes()
-        payload = get_payload_bytes(event=EVENT_TaskRequest, text=text, speaker=SPEAKER)
+        # 使用选择的speaker
+        payload = get_payload_bytes(event=EVENT_TaskRequest, text=text, speaker=selected_speaker)
         await send_event(tts_state.volc_ws, header, optional, payload)
-
         # 接收TTS响应并发送到前端
         print(f"开始接收TTS响应 [{message_id}]...")
         audio_count = 0
-
         try:
             while True:
                 raw_data = await asyncio.wait_for(
@@ -396,17 +392,13 @@ async def process_tts_task(websocket, message_id: str, text: str):
                     timeout=30
                 )
                 res = parser_response(raw_data)
-
                 print(f"收到TTS事件 [{message_id}]: {res.optional.event}")
-
                 if res.optional.event == EVENT_TTSSentenceEnd:
                     print(f"句子结束事件 [{message_id}] - 直接完成")
                     break
-
                 elif res.optional.event == EVENT_SessionFinished:
                     print(f"收到会话结束事件 [{message_id}]")
                     break
-
                 elif res.optional.event == EVENT_TTSResponse:
                     audio_count += 1
                     print(f"收到音频数据 [{message_id}] #{audio_count}，大小: {len(res.payload)}")
@@ -423,10 +415,8 @@ async def process_tts_task(websocket, message_id: str, text: str):
                     })
                 else:
                     print(f"未知TTS事件 [{message_id}]: {res.optional.event}")
-
         except asyncio.TimeoutError:
             print(f"TTS响应超时 [{message_id}]，强制结束")
-
         # 异步保存音频文件
         if tts_state.audio_data:
             file_path = await save_audio_file(
@@ -435,15 +425,15 @@ async def process_tts_task(websocket, message_id: str, text: str):
             )
             print(f"音频文件已保存 [{message_id}]: {file_path}")
 
-        # 发送完成消息，包含文件路径
+        # 发送完成消息，包含文件路径和使用的speaker
         await websocket.send_json({
             "type": "tts_audio_complete",
             "messageId": message_id,
             "audioFile": tts_state.audio_filename,
-            "audioPath": os.path.join(TEMP_AUDIO_DIR, tts_state.audio_filename) if tts_state.audio_data else None
+            "audioPath": os.path.join(TEMP_AUDIO_DIR, tts_state.audio_filename) if tts_state.audio_data else None,
+            "speaker": selected_speaker
         })
-        print(f"TTS处理完成 [{message_id}]，共发送 {audio_count} 个音频包")
-
+        print(f"TTS处理完成 [{message_id}]，共发送 {audio_count} 个音频包，使用说话人: {selected_speaker}")
     except asyncio.CancelledError:
         print(f"TTS任务被取消 [{message_id}]")
         await websocket.send_json({
@@ -474,14 +464,14 @@ async def process_tts_task(websocket, message_id: str, text: str):
 
 
 # 启动TTS文本转换
-async def handle_tts_text(websocket, message_id: str, text: str):
+async def handle_tts_text(websocket, message_id: str, text: str, speaker: str = None):
     """启动TTS文本转换"""
     # 创建新的TTS状态
+    print(speaker)
     tts_state = tts_manager.add_tts_state(websocket, message_id)
-
-    # 启动异步任务
+    # 启动异步任务，传入speaker参数
     tts_state.task = asyncio.create_task(
-        process_tts_task(websocket, message_id, text)
+        process_tts_task(websocket, message_id, text, speaker)
     )