fix legacy protocol in wav client

examples/wav_client.py

@@ -105,6 +105,7 @@ class WavFileClient:
         self.track_started = False
         self.track_ended = False
         self.send_completed = False
+        self.session_ready = False
         
         # Events log
         self.events_log = []
@@ -132,12 +133,17 @@ class WavFileClient:
         self.running = True
         self.log_event("←", "Connected!")
 
-        # Send invite command
+        # WS v1 handshake: hello -> session.start
         await self.send_command({
-            "command": "invite",
+            "type": "hello",
-            "option": {
+            "version": "v1",
-                "codec": "pcm",
+        })
-                "sampleRate": self.sample_rate
+        await self.send_command({
+            "type": "session.start",
+            "audio": {
+                "encoding": "pcm_s16le",
+                "sample_rate_hz": self.sample_rate,
+                "channels": 1
             }
         })
     
@@ -145,12 +151,12 @@ class WavFileClient:
         """Send JSON command to server."""
         if self.ws:
             await self.ws.send(json.dumps(cmd))
-            self.log_event("→", f"Command: {cmd.get('command', 'unknown')}")
+            self.log_event("→", f"Command: {cmd.get('type', 'unknown')}")
     
     async def send_hangup(self, reason: str = "Session complete") -> None:
         """Send hangup command."""
         await self.send_command({
-            "command": "hangup",
+            "type": "session.stop",
             "reason": reason
         })
     
@@ -277,51 +283,50 @@ class WavFileClient:
     
     async def _handle_event(self, event: dict) -> None:
         """Handle incoming event."""
-        event_type = event.get("event", "unknown")
+        event_type = event.get("type", "unknown")
 
-        if event_type == "answer":
+        if event_type == "hello.ack":
+            self.log_event("←", "Handshake acknowledged")
+        elif event_type == "session.started":
+            self.session_ready = True
             self.log_event("←", "Session ready!")
-        elif event_type == "speaking":
+        elif event_type == "input.speech_started":
             self.log_event("←", "Speech detected")
-        elif event_type == "silence":
+        elif event_type == "input.speech_stopped":
             self.log_event("←", "Silence detected")
-        elif event_type == "transcript":
+        elif event_type == "transcript.delta":
-            # ASR transcript (interim = asrDelta-style, final = asrFinal-style)
             text = event.get("text", "")
-            is_final = event.get("isFinal", False)
-            if is_final:
-                # Clear interim line and print final
-                print(" " * 80, end="\r")
-                self.log_event("←", f"→ You: {text}")
-            else:
-                # Interim result - show with indicator (overwrite same line, as in mic_client)
             display_text = text[:60] + "..." if len(text) > 60 else text
             print(f"  [listening] {display_text}".ljust(80), end="\r")
-        elif event_type == "ttfb":
+        elif event_type == "transcript.final":
+            text = event.get("text", "")
+            print(" " * 80, end="\r")
+            self.log_event("←", f"→ You: {text}")
+        elif event_type == "metrics.ttfb":
             latency_ms = event.get("latencyMs", 0)
             self.log_event("←", f"[TTFB] Server latency: {latency_ms}ms")
-        elif event_type == "llmResponse":
+        elif event_type == "assistant.response.delta":
             text = event.get("text", "")
-            is_final = event.get("isFinal", False)
+            if self.verbose and text:
-            if is_final:
-                self.log_event("←", f"LLM Response (final): {text[:100]}{'...' if len(text) > 100 else ''}")
-            elif self.verbose:
-                # Show streaming chunks only in verbose mode
                 self.log_event("←", f"LLM: {text}")
-        elif event_type == "trackStart":
+        elif event_type == "assistant.response.final":
+            text = event.get("text", "")
+            if text:
+                self.log_event("←", f"LLM Response (final): {text[:100]}{'...' if len(text) > 100 else ''}")
+        elif event_type == "output.audio.start":
             self.track_started = True
             self.response_start_time = time.time()
             self.waiting_for_first_audio = True
             self.log_event("←", "Bot started speaking")
-        elif event_type == "trackEnd":
+        elif event_type == "output.audio.end":
             self.track_ended = True
             self.log_event("←", "Bot finished speaking")
-        elif event_type == "interrupt":
+        elif event_type == "response.interrupted":
             self.log_event("←", "Bot interrupted!")
         elif event_type == "error":
-            self.log_event("!", f"Error: {event.get('error')}")
+            self.log_event("!", f"Error: {event.get('message')}")
-        elif event_type == "hangup":
+        elif event_type == "session.stopped":
-            self.log_event("←", f"Hangup: {event.get('reason')}")
+            self.log_event("←", f"Session stopped: {event.get('reason')}")
             self.running = False
         else:
             self.log_event("←", f"Event: {event_type}")
@@ -359,12 +364,16 @@ class WavFileClient:
             # Connect to server
             await self.connect()
             
-            # Wait for answer
-            await asyncio.sleep(0.5)
-            
             # Start receiver task
             receiver_task = asyncio.create_task(self.receiver())
 
+            # Wait for session.started before streaming audio
+            ready_start = time.time()
+            while self.running and not self.session_ready:
+                if time.time() - ready_start > 8.0:
+                    raise TimeoutError("Timeout waiting for session.started")
+                await asyncio.sleep(0.05)
+            
             # Send audio
             await self.audio_sender(audio_data)