adjust server

app.py

@@ -1,4 +1,4 @@
-from fastapi import FastAPI, WebSocket
+from fastapi import FastAPI, WebSocket, WebSocketDisconnect
 from contextlib import asynccontextmanager
 import uvicorn
 from whisper_live.server import TranscriptionServer
@@ -10,14 +10,14 @@ import numpy as np
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-# Initialize the transcription server
+# Initialize the transcription server once
 transcription_server = TranscriptionServer()
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    # Any startup configuration can go here
+    # Setup if needed
     yield
-    # Cleanup can go here if needed
+    # Cleanup if needed
 
 app = FastAPI(
     title="Whisper Live Server",
@@ -28,9 +28,6 @@ app = FastAPI(
 
 @app.get("/")
 async def root():
-    """
-    Root endpoint that returns a welcome message
-    """
     return {
         "message": "Welcome to Whisper Live Server",
         "status": "running",
@@ -38,64 +35,67 @@ async def root():
         "health_endpoint": "/health"
     }
 
+@app.get("/health")
+def health_check():
+    return {"status": "healthy"}
+
+
 @app.websocket("/ws")
 async def websocket_endpoint(websocket: WebSocket):
     await websocket.accept()
     client_uid = None
+    config = {}
     
     try:
-        # Receive initial configuration
+        # 1. Receive config from client
         config = await websocket.receive_json()
         client_uid = config.get("uid")
-        
+
         if not client_uid:
             await websocket.close(code=4000, reason="No client UID provided")
             return
-            
-        # Send ready message
+
+        logger.info(f"Client connected: {client_uid} | Config: {config}")
+
+        # 2. Confirm server readiness
         await websocket.send_json({
             "uid": client_uid,
             "message": "SERVER_READY",
             "backend": "faster_whisper"
         })
-        
-        # Start the transcription server
-        transcription_server.run(
-            host="0.0.0.0",
-            port=7860,
-            backend="faster_whisper",
-            single_model=True,
+
+        # 3. Per-client session configuration
+        session = transcription_server.create_session(
+            model=config.get("model", "tiny"),
             language=config.get("language", "de"),
             task=config.get("task", "transcribe"),
-            model=config.get("model", "tiny"),
             use_vad=config.get("use_vad", True)
         )
-        
-        # Handle incoming audio data
+
+        # 4. Start processing audio stream
         while True:
             data = await websocket.receive()
-            
+
             if data["type"] == "websocket.disconnect":
                 break
-                
-            if data["type"] == "websocket.receive":
-                if isinstance(data.get("bytes"), bytes):
-                    # Process binary audio data
-                    audio_data = np.frombuffer(data["bytes"], dtype=np.float32)
-                    # Process audio data through transcription server
-                    segments = transcription_server.process_audio(audio_data)
-                    
-                    if segments:
-                        await websocket.send_json({
-                            "uid": client_uid,
-                            "segments": segments
-                        })
-                elif isinstance(data.get("text"), str):
-                    # Handle text messages (like END_OF_AUDIO)
-                    msg = data["text"]
-                    if msg == "END_OF_AUDIO":
-                        break
-                        
+
+            if "bytes" in data:
+                audio_data = np.frombuffer(data["bytes"], dtype=np.float32)
+                segments = session.process_audio(audio_data)
+
+                if segments:
+                    await websocket.send_json({
+                        "uid": client_uid,
+                        "segments": segments
+                    })
+
+            elif "text" in data:
+                if data["text"] == "END_OF_AUDIO":
+                    logger.info(f"Client {client_uid} ended stream.")
+                    break
+
+    except WebSocketDisconnect:
+        logger.warning(f"WebSocket disconnected: {client_uid}")
     except Exception as e:
         logger.error(f"WebSocket error: {e}")
         if client_uid:
@@ -107,17 +107,7 @@ async def websocket_endpoint(websocket: WebSocket):
             except:
                 pass
     finally:
-        try:
-            await websocket.close()
-        except:
-            pass
-
-@app.get("/health")
-def health_check():
-    """
-    Health check endpoint to verify the server is running
-    """
-    return {"status": "healthy"}
+        await websocket.close()
 
 if __name__ == "__main__":
-    uvicorn.run(app, host="0.0.0.0", port=7860) 
+    uvicorn.run(app, host="0.0.0.0", port=7860)