Update websocket_handler.py

websocket_handler.py

@@ -455,38 +455,38 @@ async def handle_control_message(websocket: WebSocket, session: RealtimeSession,
             }
         })
         
-    elif action == "end_session":
-        # Clean up and close
-        await session.cleanup()
-        await websocket.close()
-        
-    elif action == "interrupt":
-        # Handle explicit interrupt
-        await session.handle_barge_in()
-        await websocket.send_json({
-            "type": "control",
-            "action": "interrupt_acknowledged"
-        })
-        
-    elif action == "reset":
-        # Reset conversation state
-        await session.reset_for_new_utterance()
-        await session.change_state(ConversationState.IDLE)
-        await websocket.send_json({
-            "type": "state_change",
-            "from": session.state.value,
-            "to": "idle"
-        })
-        
-    elif action == "audio_ended":
-        # Audio playback ended on client
-        if session.state == ConversationState.PLAYING_AUDIO:
-            await session.change_state(ConversationState.IDLE)
-            await websocket.send_json({
-                "type": "state_change",
-                "from": "playing_audio",
-                "to": "idle"
-            })
+        # Send welcome message and TTS if available
+        chat_history = session.session.chat_history
+        if chat_history and len(chat_history) > 0:
+            last_message = chat_history[-1]
+            if last_message["role"] == "assistant":
+                welcome_text = last_message["content"]
+                
+                # Send text message
+                await websocket.send_json({
+                    "type": "assistant_response",
+                    "text": welcome_text
+                })
+                
+                # Generate TTS if enabled
+                tts_provider = TTSFactory.create_provider()
+                if tts_provider:
+                    await session.change_state(ConversationState.PROCESSING_TTS)
+                    await websocket.send_json({
+                        "type": "state_change",
+                        "from": "idle",
+                        "to": "processing_tts"
+                    })
+                    
+                    # Generate and stream TTS
+                    tts_task = session.barge_in_handler.start_tts_task(
+                        generate_and_stream_tts(websocket, session, tts_provider, welcome_text)
+                    )
+                    
+                    try:
+                        await tts_task
+                    except asyncio.CancelledError:
+                        log_info("Welcome TTS cancelled", session_id=session.session.session_id)
 
 
 # ========================= PROCESSING FUNCTIONS =========================
@@ -620,6 +620,46 @@ async def generate_and_stream_tts(
             chunk = audio_data[i:i + chunk_size]
             chunk_index = i // chunk_size
             
+            await websocket.send_json({
+                "type": "tts_audio",
+                "data": base64.b64encode(chunk).decode('utf-8'),
+                "chunk_index": chunk_index,
+                "total_chunks": total_chunks,
+                "is_last": chunk_index == total_chunks - 1,
+                "mime_type": "audio/mpeg"  # MP3 format for ElevenLabs
+            })
+            
+            # Small delay to prevent overwhelming the client
+            await asyncio.sleep(0.01)
+        
+        # Send state back to idle after completion
+        await session.change_state(ConversationState.IDLE)
+        await websocket.send_json({
+            "type": "state_change",
+            "from": "playing_audio",
+            "to": "idle"
+        })
+        
+        log_info(
+            f"TTS streaming completed",
+            session_id=session.session.session_id,
+            text_length=len(text),
+            audio_size=len(audio_data)
+        )
+        
+    except asyncio.CancelledError:
+        log_info("TTS streaming cancelled", session_id=session.session.session_id)
+        raise
+    except Exception as e:
+        log_error(
+            f"TTS generation error",
+            error=str(e),
+            session_id=session.session.session_id
+        )
+        await websocket.send_json({
+            "type": "error",
+            "message": f"TTS error: {str(e)}"
+        })
             await websocket.send_json({
                 "type": "tts_audio",
                 "data": base64.b64encode(chunk).decode('utf-8'),