feat:将transcriber作为环境变量配置

.env.example

@@ -25,4 +25,8 @@ QWEN_API_BASE_URL=
 QWEN_MODEL=
 MODEl_PROVIDER= #如果不是openai 请修改 deepseek/qwen
 # FFMPEG 配置
-FFMPEG_BIN_PATH=
+FFMPEG_BIN_PATH=
+
+# transcriber 相关配置
+TRANSCRIBER_TYPE=fast-whisper # fast-whisper/bcut/kuaishou
+WHISPER_MODEL_SIZE=base

backend/.env.example

@@ -21,3 +21,7 @@ QWEN_API_KEY=
 QWEN_API_BASE_URL=
 QWEN_MODEL=
 MODEl_PROVIDER= #如果不是openai 请修改 deepseek/qwen
+
+# transcriber 相关配置
+TRANSCRIBER_TYPE=fast-whisper # fast-whisper/bcut/kuaishou
+WHISPER_MODEL_SIZE=base

backend/app/services/note.py

@@ -43,7 +43,7 @@ class NoteGenerator:
     def __init__(self):
         self.model_size: str = 'base'
         self.device: Union[str, None] = None
-        self.transcriber_type = 'fast-whisper'
+        self.transcriber_type = os.getenv('TRANSCRIBER_TYPE','fast-whisper')
         self.transcriber = self.get_transcriber()
         # TODO 需要更换为可调节
 
@@ -91,7 +91,7 @@ class NoteGenerator:
         '''
         if self.transcriber_type == 'fast-whisper':
             logger.info("使用Whisper")
-            return get_transcriber()
+            return get_transcriber(transcriber_type='fast-whisper')
         else:
             logger.warning("不支持的转义器")
             raise ValueError(f"不支持的转义器：{self.transcriber}")

backend/app/transcriber/transcriber_provider.py

@@ -1,3 +1,5 @@
+import os
+
 from app.transcriber.whisper import WhisperTranscriber
 from app.transcriber.bcut import BcutTranscriber
 from app.transcriber.kuaishou import KuaishouTranscriber
@@ -49,12 +51,12 @@ def get_kuaishou_transcriber():
             raise
     return _transcribers['kuaishou']
 
-def get_transcriber(transcriber_type="whisper", model_size="base", device="cuda"):
+def get_transcriber(transcriber_type="fast-whisper", model_size="base", device="cuda"):
     """
     获取指定类型的转录器实例
     
     参数:
-        transcriber_type: 转录器类型，支持 "whisper", "bcut", "kuaishou"
+        transcriber_type: 转录器类型，支持 "fast-whisper", "bcut", "kuaishou"
         model_size: 模型大小，whisper 特有参数
         device: 设备类型，whisper 特有参数
     
@@ -62,13 +64,14 @@ def get_transcriber(transcriber_type="whisper", model_size="base", device="cuda"
         对应类型的转录器实例
     """
     logger.info(f'获取转录器，类型: {transcriber_type}')
-    
+    if transcriber_type == "fast-whisper":
-    if transcriber_type == "whisper":
+        whisper_model_size = os.environ.get("WHISPER_MODEL_SIZE",model_size)
-        return get_whisper_transcriber(model_size, device)
+        return get_whisper_transcriber(whisper_model_size, device=device)
     elif transcriber_type == "bcut":
         return get_bcut_transcriber()
     elif transcriber_type == "kuaishou":
         return get_kuaishou_transcriber()
     else:
         logger.warning(f'未知转录器类型 "{transcriber_type}"，使用默认 whisper')
-        return get_whisper_transcriber(model_size, device)
+        whisper_model_size = os.environ.get("WHISPER_MODEL_SIZE",model_size)
+        return get_whisper_transcriber(whisper_model_size, device)

backend/main.py

@@ -34,7 +34,7 @@ async def startup_event():
 async def startup_event():
     register_handler()
     ensure_ffmpeg_or_raise()
-    get_transcriber()
+    get_transcriber(transcriber_type=os.getenv("TRANSCRIBER_TYPE","fast-whisper"))
     init_video_task_table()
 
 if __name__ == "__main__":