Merge pull request #2114 from xinnan-tech/paddlespeech

update：更新paddlespeechtts供应器

docs/paddlespeech-deploy.md

@@ -75,7 +75,7 @@ TTS:
       sample_rate: 24000  # 采样率 [websocket默认24000，http默认0 自动选择]
       speed: 1.0  # 语速，1.0 表示正常语速，>1 表示加快，<1 表示减慢
       volume: 1.0  # 音量，1.0 表示正常音量，>1 表示增大，<1 表示减小
-      save_path: ./streaming_tts.wav  # 服务器生成的语音文件保存路径
+      save_path:   # 保存路径
 ```
 ### 3.启动xiaozhi服务
 ```py

main/xiaozhi-server/config.yaml

@@ -895,7 +895,7 @@ TTS:
     sample_rate: 24000  # 采样率 [websocket默认24000，http默认0 自动选择]
     speed: 1.0  # 语速，1.0 表示正常语速，>1 表示加快，<1 表示减慢
     volume: 1.0  # 音量，1.0 表示正常音量，>1 表示增大，<1 表示减小
-    save_path: ./streaming_tts.wav  # 服务器生成的语音文件保存路径
+    save_path:   # 保存路径
   IndexStreamTTS:
     # 基于Index-TTS-vLLM项目的TTS接口服务
     # 参照教程：https://github.com/Ksuriuri/index-tts-vllm/blob/master/README.md

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

@@ -8,6 +8,7 @@ import wave
 import websockets
 from core.providers.tts.base import TTSProviderBase
 from config.logger import setup_logging
+from datetime import datetime
 
 TAG = __name__
 logger = setup_logging()
@@ -18,11 +19,12 @@ class TTSProvider(TTSProviderBase):
         super().__init__(config, delete_audio_file)
         self.url = config.get("url", "ws://192.168.1.10:8092/paddlespeech/tts/streaming")
         self.protocol = config.get("protocol", "websocket")
+        
         if config.get("private_voice"):
             self.spk_id = int(config.get("private_voice"))
         else:
-            self.spk_id = int(config.get("spk_id", "0"))  
-        
+            self.spk_id = int(config.get("spk_id", "0"))
+            
         sample_rate = config.get("sample_rate", 24000)
         self.sample_rate = float(sample_rate) if sample_rate else 24000
         
@@ -32,7 +34,21 @@ class TTSProvider(TTSProviderBase):
         volume = config.get("volume", 1.0)
         self.volume = float(volume) if volume else 1.0
         
-        self.save_path = config.get("save_path", "./streaming_tts.wav")
+        self.delete_audio_file = config.get("delete_audio", True)
+        if not self.delete_audio_file:
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            save_path = config.get("save_path")
+            if save_path:
+                if not save_path.endswith('.wav'):
+                    save_path = f"{save_path}_{timestamp}.wav"
+                else:
+                    other_path = save_path[:-4]
+                    save_path = f"{other_path}_{timestamp}.wav"
+                self.save_path = save_path
+            else:
+                self.save_path = f"./streaming_tts_{timestamp}.wav"
+        else:
+            self.save_path = None
 
     async def pcm_to_wav(self, pcm_data: bytes, sample_rate: int = 24000, num_channels: int = 1,
                          bits_per_sample: int = 16) -> bytes:
@@ -151,6 +167,12 @@ class TTSProvider(TTSProviderBase):
                 # 接收结束响应避免服务抛出异常
                 await ws.recv()
 
+                # 根据配置决定是否保存文件
+                if not self.delete_audio_file and self.save_path:
+                    with open(self.save_path, "wb") as f:
+                        f.write(wav_data)
+                    logger.bind(tag=TAG).info(f"音频文件已保存到: {self.save_path}")
+                
                 # 返回或保存音频数据
                 if output_file:
                     with open(output_file, "wb") as file_to_save:
@@ -159,4 +181,4 @@ class TTSProvider(TTSProviderBase):
                     return wav_data
 
         except Exception as e:
-            raise Exception(f"Error during TTS WebSocket request: {e} while processing text: {text}")
+            raise Exception(f"Error during TTS WebSocket request: {e} while processing text: {text}")