Update websocket_handler.py

websocket_handler.py

@@ -473,94 +473,74 @@ async def handle_audio_chunk(websocket: WebSocket, session: RealtimeSession, mes
         # Check silence
         silence_duration = session.silence_detector.update(decoded_audio)
         
-        # Stream to STT if available
-        if session.stt_manager and session.state == ConversationState.LISTENING:
-            # Ensure streaming is active
-            if not session.is_streaming:
-                log_warning(f"⚠️ STT manager exists but streaming not active", session_id=session.session.session_id)
-                # Try to restart streaming
-                stt_initialized = await session.initialize_stt()
-                if not stt_initialized:
-                    await websocket.send_json({
-                        "type": "error",
-                        "error_type": "stt_error",
-                        "message": "STT streaming not available"
-                    })
-                    return
+    # Stream to STT if available
+    if session.stt_manager and session.state == ConversationState.LISTENING:
+        # Ensure streaming is active
+        if not session.is_streaming:
+            log_warning(f"⚠️ STT manager exists but streaming not active", session_id=session.session.session_id)
+            # Try to restart streaming
+            stt_initialized = await session.initialize_stt()
+            if not stt_initialized:
+                await websocket.send_json({
+                    "type": "error",
+                    "error_type": "stt_error",
+                    "message": "STT streaming not available"
+                })
+                return
+        
+        try:
+            # Chunk counter - sadece önemli milestone'larda logla
+            if not hasattr(session, 'chunk_counter'):
+                session.chunk_counter = 0
+            session.chunk_counter += 1
             
-            try:
-                # Chunk counter - sadece önemli milestone'larda logla
-                if not hasattr(session, 'chunk_counter'):
-                    session.chunk_counter = 0
-                session.chunk_counter += 1
-                
-                if session.chunk_counter == 1:
-                    log_info(f"🎤 Started streaming audio to STT", session_id=session.session.session_id)
-                elif session.chunk_counter % 100 == 0:
-                    log_info(f"📊 Sent {session.chunk_counter} chunks to STT so far...", session_id=session.session.session_id)
+            if session.chunk_counter == 1:
+                log_info(f"🎤 Started streaming audio to STT", session_id=session.session.session_id)
+            elif session.chunk_counter % 100 == 0:
+                log_info(f"📊 Sent {session.chunk_counter} chunks to STT so far...", session_id=session.session.session_id)
+            
+            # STT'ye gönder ve sonuçları bekle
+            async for result in session.stt_manager.stream_audio(decoded_audio):
+                # Sadece anlamlı sonuçları logla
+                if result.text.strip():  # Boş olmayan text varsa
+                    log_info(f"🎤 STT: '{result.text}' (final: {result.is_final})", session_id=session.session.session_id)
                 
-                # STT'ye gönder ve sonuçları bekle
-                result_received = False
+                # Send transcription updates
+                await websocket.send_json({
+                    "type": "transcription",
+                    "text": result.text,
+                    "is_final": result.is_final,
+                    "confidence": result.confidence
+                })
                 
-                async for result in session.stt_manager.stream_audio(decoded_audio):
-                    result_received = True
-                    
-                    # Sadece anlamlı sonuçları logla
-                    if result.text.strip():  # Boş olmayan text varsa
-                        log_info(f"🎤 STT: '{result.text}' (final: {result.is_final})", session_id=session.session.session_id)
-                    
-                    # Send transcription updates
-                    await websocket.send_json({
-                        "type": "transcription",
-                        "text": result.text,
-                        "is_final": result.is_final,
-                        "confidence": result.confidence
-                    })
+                if result.is_final:
+                    session.current_transcription = result.text
+                    log_info(f"✅ FINAL TRANSCRIPTION: '{result.text}'", session_id=session.session.session_id)
                     
-                    if result.is_final:
-                        session.current_transcription = result.text
-                        log_info(f"✅ FINAL TRANSCRIPTION: '{result.text}'", session_id=session.session.session_id)
+                    # Final transcription geldiğinde hemen işle
+                    if session.current_transcription:
+                        # State'i değiştir ve user input'u işle
+                        await session.change_state(ConversationState.PROCESSING_STT)
+                        await websocket.send_json({
+                            "type": "state_change",
+                            "from": "listening",
+                            "to": "processing_stt"
+                        })
                         
-                        # Final transcription geldiğinde hemen işle
-                        if session.current_transcription:
-                            # State'i değiştir ve user input'u işle
-                            await session.change_state(ConversationState.PROCESSING_STT)
-                            await websocket.send_json({
-                                "type": "state_change",
-                                "from": "listening",
-                                "to": "processing_stt"
-                            })
-                            
-                            # Process user input
-                            await process_user_input(websocket, session)
-                            
-                            # STT'den final result geldiğinde audio buffer'ı ve transcription'ı resetle
-                            await session.reset_for_new_utterance()
-                            return  # Bu audio chunk için işlem tamamlandı
-                
-                # Her 200 chunk'ta bir result gelmiyorsa uyar
-                if not result_received and session.chunk_counter % 200 == 0:
-                    log_warning(f"⚠️ No STT results after {session.chunk_counter} chunks", session_id=session.session.session_id)
+                        # Process user input
+                        await process_user_input(websocket, session)
                         
-            except Exception as e:
-                log_error(f"❌ STT streaming error", error=str(e), traceback=traceback.format_exc(), session_id=session.session.session_id)
-                await websocket.send_json({
-                    "type": "error",
-                    "error_type": "stt_error",
-                    "message": f"STT error: {str(e)}"
-                })
-            
-    except Exception as e:
-        log_error(
-            f"❌ Audio chunk handling error",
-            error=str(e),
-            traceback=traceback.format_exc(),
-            session_id=session.session.session_id
-        )
-        await websocket.send_json({
-            "type": "error",
-            "message": f"Audio processing error: {str(e)}"
-        })
+                        # STT'den final result geldiğinde audio buffer'ı ve transcription'ı resetle
+                        await session.reset_for_new_utterance()
+                        return  # Bu audio chunk için işlem tamamlandı
+                        
+        except Exception as e:
+            log_error(f"❌ STT streaming error", error=str(e), traceback=traceback.format_exc(), session_id=session.session.session_id)
+            await websocket.send_json({
+                "type": "error",
+                "error_type": "stt_error",
+                "message": f"STT error: {str(e)}"
+            })
         
 async def handle_control_message(websocket: WebSocket, session: RealtimeSession, message: Dict[str, Any]):
     """Handle control messages"""