add：双向流tts结构改造，火山双向tts语音流式输入输出

main/xiaozhi-server/core/providers/tts/base.py

@@ -50,21 +50,9 @@ class TTSProviderBase(ABC):
         self.executor = ThreadPoolExecutor(max_workers=self.max_workers)
 
     async def open_audio_channels(self):
-        pass
-
-    async def reset(self):
-        try:
-            logger.bind(tag=TAG).info("说明开始了新的对话，重建tts监听")
-            await self.stop_listen_resource()
-            self.tts_text_queue = queue.Queue()
-            self.tts_audio_queue = queue.Queue()
-            # 启动tts_text_queue监听线程
-            self.stop_event.clear()
-            tts_priority = threading.Thread(target=self._tts_text_priority_thread, daemon=True)
-            tts_priority.start()
-        except Exception as e:
-            logger.bind(tag=TAG).error(f"Failed to process TTS text: {e}")
-            traceback.print_exc()
+        # 启动tts_text_queue监听线程
+        tts_priority = threading.Thread(target=self._tts_text_priority_thread, daemon=True)
+        tts_priority.start()
 
     async def stop_listen_resource(self):
         """资源清理方法"""

main/xiaozhi-server/core/providers/tts/fishspeech.py

@@ -180,8 +180,8 @@ class TTSProvider(TTSProviderBase):
                         if len(chunk_total) % 2 == 0 and chunk_total[-2:] == b'\x00\x00':
                             audio = self._get_audio_from_tts(chunk_total)
                             audio_raw = audio_raw + audio.raw_data
-                            # 长度凑够2贞开始发送，60ms*4=240ms
-                            if len(audio_raw) >= 7680:
+                            # 长度凑够2贞开始发送，60ms*2=120ms
+                            if len(audio_raw) >= 3840:
                                 opus_datas = self.wav_to_opus_data_audio_raw(audio_raw)
                                 if index == 0:
                                     yield TTSMessageDTO(u_id=u_id, msg_type=MsgType.TTS_TEXT_RESPONSE,

main/xiaozhi-server/core/providers/tts/huoshan.py

@@ -152,7 +152,7 @@ class TTSProvider(TTSProviderBase):
         self.tts_text = ""
 
     async def open_audio_channels(self):
-        self.loop_tts = asyncio.new_event_loop()
+        await super().open_audio_channels()
         ws_header = {
             "X-Api-App-Key": self.appId,
             "X-Api-Access-Key": self.access_token,
@@ -316,6 +316,7 @@ class TTSProvider(TTSProviderBase):
 
     async def close(self):
         """资源清理方法"""
+        await self.finish_connection()
         await self.ws.close()
 
     async def text_to_speak(self, u_id, text, is_last_text=False, is_first_text=False):