Audio Client

python examples/minimal_server.py

import websockets

async with websockets.connect(
    "ws://localhost:8000/ws",
    extra_headers={"X-API-Key": "demo-api-key"}
) as websocket:
    # Connection established

start_msg = {
    "type": "client_audio_start",
    "session_id": session_id,
    "data": {
        "sample_rate": 16000,
        "encoding": "PCM16"
    }
}
await websocket.send(json.dumps(start_msg))

# Capture audio from microphone
audio_data = capture_audio_chunk()

# Base64 encode
b64_audio = base64.b64encode(audio_data).decode('utf-8')

# Send chunk
chunk_msg = {
    "type": "client_audio_chunk",
    "session_id": session_id,
    "data": b64_audio
}
await websocket.send(json.dumps(chunk_msg))

response = await websocket.recv()
message = json.loads(response)

if message["type"] == "server_stt":
    print(f"STT: {message['data']['text']}")
elif message["type"] == "server_text":
    print(f"Agent: {message['data']['text']}")

python examples/minimal_server.py

python examples/clients/send_audio_client.py

python examples/clients/send_audio_client.py \
    --server ws://localhost:8000/ws \
    --api-key demo-api-key

🎤 Recording started (16000Hz, Mono)...
📤 Sent audio start message
📤 Streaming audio... (Press Ctrl+C to stop)

STT: Hello, how are you?
Agent: I'm doing well, thank you! How can I help you today?

🛑 Recording stopped.

1. Connect to WebSocket
   ↓
2. Send client_audio_start
   ↓
3. Stream client_audio_chunk (continuous)
   ↓
4. Receive server_stt (transcriptions)
   ↓
5. Receive server_text (agent responses)
   ↓
6. Send client_audio_end (when done)