Update app.py

app.py

@@ -5,11 +5,11 @@ from google import genai
 
 app = Quart(__name__)
 
-# Ensure your HF Space has GEMINI_API_KEY set in its secrets/environment variables
-client = genai.Client()
+# Force v1beta version specifically for Multimodal Live API features
+client = genai.Client(http_options={'api_version': 'v1beta'})
 
-# Note: Using gemini-live-2.5-flash-preview-native-audio-09-2025 as it is the most free for the Live SDK currently
-MODEL = "gemini-live-2.5-flash-preview-native-audio-09-2025" 
+# This is the standard model ID for bidirectional streaming
+MODEL = "gemini-2.0-flash-exp" 
 
 VOICE_MODES = {
     'Zephyr': 'Zephyr', # Default / Balanced
@@ -28,6 +28,15 @@ async def index():
         "supported_voices": list(VOICE_MODES.keys())
     }
 
+@app.route('/models')
+async def list_models():
+    """Use this to see exactly which models your key has access to if it fails."""
+    try:
+        models = client.models.list()
+        return {"available_models": [m.name for m in models]}
+    except Exception as e:
+        return {"error": str(e)}
+
 @app.websocket('/stream')
 async def ws_stream():
     """
@@ -37,7 +46,6 @@ async def ws_stream():
     requested_voice = websocket.args.get("voice", "Zephyr")
     voice_name = VOICE_MODES.get(requested_voice, "Zephyr")
 
-    # Using a dictionary for config prevents AttributeError on specific SDK versions
     config = {
         "response_modalities": ["AUDIO"],
         "speech_config": {
@@ -56,17 +64,14 @@ async def ws_stream():
     print(f"Connecting to Gemini Live API with voice: {voice_name}...")
     
     try:
-        # Pass the dictionary directly to the config parameter
         async with client.aio.live.connect(model=MODEL, config=config) as session:
             print("Live session established.")
 
-            # Task 1: Stream audio from Client (Termux) -> Gemini
             async def client_to_gemini():
                 try:
                     while True:
                         data = await websocket.receive()
                         if isinstance(data, bytes):
-                            # Sending 16kHz PCM data from client to Gemini
                             await session.send(
                                 input={"data": data, "mime_type": "audio/pcm;rate=16000"}
                             )
@@ -75,7 +80,6 @@ async def ws_stream():
                 except Exception as e:
                     print(f"Error reading from client: {e}")
 
-            # Task 2: Stream audio from Gemini -> Client (Termux)
             async def gemini_to_client():
                 try:
                     async for message in session.receive():
@@ -88,14 +92,12 @@ async def ws_stream():
                             if model_turn:
                                 for part in model_turn.parts:
                                     if part.inline_data and part.inline_data.data:
-                                        # Sending 24kHz PCM data back to client
                                         await websocket.send(part.inline_data.data)
                 except asyncio.CancelledError:
                     pass
                 except Exception as e:
                     print(f"Error receiving from Gemini: {e}")
 
-            # Run both streaming directions concurrently
             task1 = asyncio.create_task(client_to_gemini())
             task2 = asyncio.create_task(gemini_to_client())
             
@@ -111,5 +113,4 @@ async def ws_stream():
         print(f"Connection failed: {e}")
 
 if __name__ == "__main__":
-    # HF Spaces standard port is 7860
     app.run(host="0.0.0.0", port=7860)