更新paddlespeechtts供应器

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

@@ -1,4 +1,3 @@
-
 import asyncio
 import json
 import base64
@@ -9,6 +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()
@@ -19,11 +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
         
@@ -33,7 +34,12 @@ 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", "None")
+        self.delete_audio_file = config.get("delete_audio_file", True)
+        if not self.delete_audio_file:
+            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+            self.save_path = config.get("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:
@@ -56,22 +62,22 @@ class TTSProvider(TTSProviderBase):
 
         return wav_io.getvalue()
 
-    async def text_to_speak(self, text, output_file=None):
+    async def text_to_speak(self, text, output_file):
         if self.protocol == "websocket":
-            return await self.text_streaming(text)
+            return await self.text_streaming(text, output_file)
         elif self.protocol == "http":
-            return await self.text(text)
+            return await self.text(text, output_file)
         else:
             raise ValueError("Unsupported protocol. Please use 'websocket' or 'http'.")
 
-    async def text(self, text):
+    async def text(self, text, output_file):
         request_json = {
             "text": text,
             "spk_id": self.spk_id,
             "speed": self.speed,
             "volume": self.volume,
             "sample_rate": self.sample_rate,
-            "save_path": None,
+            "save_path": self.save_path
         }
 
         try:
@@ -82,7 +88,11 @@ class TTSProvider(TTSProviderBase):
                         if resp_json.get("success"):
                             data = resp_json["result"]
                             audio_bytes = base64.b64decode(data["audio"])
-                            return audio_bytes
+                            if output_file:
+                                with open(output_file, "wb") as file_to_save:
+                                    file_to_save.write(audio_bytes)
+                            else:
+                                return audio_bytes
                         else:
                             raise Exception(
                                 f"Error: {resp_json.get('message', 'Unknown error')} while processing text: {text}")
@@ -92,7 +102,7 @@ class TTSProvider(TTSProviderBase):
         except Exception as e:
             raise Exception(f"Error during TTS HTTP request: {e} while processing text: {text}")
 
-    async def text_streaming(self, text):
+    async def text_streaming(self, text, output_file):
         try:
             # 使用 websockets 异步连接到 WebSocket 服务器
             async with websockets.connect(self.url) as ws:
@@ -148,8 +158,18 @@ class TTSProvider(TTSProviderBase):
                 # 接收结束响应避免服务抛出异常
                 await ws.recv()
 
-                # 直接返回音频数据
-                return wav_data
+                # 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:
+                        file_to_save.write(wav_data)
+                else:
+                    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}")