1
import pyaudio
2
import websocket
3
import json
4
import threading
5
import time
6
from urllib.parse import urlencode
7
from datetime import datetime
8
9
# --- Configuration ---
10
YOUR_API_KEY = "YOUR-API-KEY"  # Replace with your actual API key
11
12
CONNECTION_PARAMS = {
13
    "sample_rate": 16000,
14
    "speech_model": "u3-rt-pro",
15
    "keyterms_prompt": json.dumps(["Keanu Reeves", "AssemblyAI", "Universal-2"])
16
}
17
API_ENDPOINT_BASE_URL = "wss://streaming.assemblyai.com/v3/ws"
18
API_ENDPOINT = f"{API_ENDPOINT_BASE_URL}?{urlencode(CONNECTION_PARAMS)}"
19
20
# Audio Configuration
21
FRAMES_PER_BUFFER = 800  # 50ms of audio (0.05s * 16000Hz)
22
SAMPLE_RATE = CONNECTION_PARAMS["sample_rate"]
23
CHANNELS = 1
24
FORMAT = pyaudio.paInt16
25
26
# Global variables for audio stream and websocket
27
audio = None
28
stream = None
29
ws_app = None
30
audio_thread = None
31
stop_event = threading.Event()  # To signal the audio thread to stop
32
33
# --- WebSocket Event Handlers ---
34
35
36
def on_open(ws):
37
    """Called when the WebSocket connection is established."""
38
    print("WebSocket connection opened.")
39
    print(f"Connected to: {API_ENDPOINT}")
40
41
    # Start sending audio data in a separate thread
42
    def stream_audio():
43
        global stream
44
        print("Starting audio streaming...")
45
        while not stop_event.is_set():
46
            try:
47
                audio_data = stream.read(FRAMES_PER_BUFFER, exception_on_overflow=False)
48
49
                # Send audio data as binary message
50
                ws.send(audio_data, websocket.ABNF.OPCODE_BINARY)
51
            except Exception as e:
52
                print(f"Error streaming audio: {e}")
53
                # If stream read fails, likely means it's closed, stop the loop
54
                break
55
        print("Audio streaming stopped.")
56
57
    global audio_thread
58
    audio_thread = threading.Thread(target=stream_audio)
59
    audio_thread.daemon = (
60
        True  # Allow main thread to exit even if this thread is running
61
    )
62
    audio_thread.start()
63
64
def on_message(ws, message):
65
    try:
66
        data = json.loads(message)
67
        msg_type = data.get('type')
68
69
        if msg_type == "Begin":
70
            session_id = data.get('id')
71
            expires_at = data.get('expires_at')
72
            print(f"\nSession began: ID={session_id}, ExpiresAt={datetime.fromtimestamp(expires_at)}")
73
        elif msg_type == "Turn":
74
            transcript = data.get('transcript', '')
75
            if data.get('end_of_turn'):
76
                print('\r' + ' ' * 80 + '\r', end='')
77
                print(transcript)
78
            else:
79
                print(f"\r{transcript}", end='')
80
        elif msg_type == "Termination":
81
            audio_duration = data.get('audio_duration_seconds', 0)
82
            session_duration = data.get('session_duration_seconds', 0)
83
            print(f"\nSession Terminated: Audio Duration={audio_duration}s, Session Duration={session_duration}s")
84
    except json.JSONDecodeError as e:
85
        print(f"Error decoding message: {e}")
86
    except Exception as e:
87
        print(f"Error handling message: {e}")
88
89
def on_error(ws, error):
90
    """Called when a WebSocket error occurs."""
91
    print(f"\nWebSocket Error: {error}")
92
    # Attempt to signal stop on error
93
    stop_event.set()
94
95
96
def on_close(ws, close_status_code, close_msg):
97
    """Called when the WebSocket connection is closed."""
98
    print(f"\nWebSocket Disconnected: Status={close_status_code}, Msg={close_msg}")
99
100
    # Ensure audio resources are released
101
    global stream, audio
102
    stop_event.set()  # Signal audio thread just in case it's still running
103
104
    if stream:
105
        if stream.is_active():
106
            stream.stop_stream()
107
        stream.close()
108
        stream = None
109
    if audio:
110
        audio.terminate()
111
        audio = None
112
    # Try to join the audio thread to ensure clean exit
113
    if audio_thread and audio_thread.is_alive():
114
        audio_thread.join(timeout=1.0)
115
116
117
# --- Main Execution ---
118
def run():
119
    global audio, stream, ws_app
120
121
    # Initialize PyAudio
122
    audio = pyaudio.PyAudio()
123
124
    # Open microphone stream
125
    try:
126
        stream = audio.open(
127
            input=True,
128
            frames_per_buffer=FRAMES_PER_BUFFER,
129
            channels=CHANNELS,
130
            format=FORMAT,
131
            rate=SAMPLE_RATE,
132
        )
133
        print("Microphone stream opened successfully.")
134
        print("Speak into your microphone. Press Ctrl+C to stop.")
135
    except Exception as e:
136
        print(f"Error opening microphone stream: {e}")
137
        if audio:
138
            audio.terminate()
139
        return  # Exit if microphone cannot be opened
140
141
    # Create WebSocketApp
142
    ws_app = websocket.WebSocketApp(
143
        API_ENDPOINT,
144
        header={"Authorization": YOUR_API_KEY},
145
        on_open=on_open,
146
        on_message=on_message,
147
        on_error=on_error,
148
        on_close=on_close,
149
    )
150
151
    # Run WebSocketApp in a separate thread to allow main thread to catch KeyboardInterrupt
152
    ws_thread = threading.Thread(target=ws_app.run_forever)
153
    ws_thread.daemon = True
154
    ws_thread.start()
155
156
    try:
157
        # Keep main thread alive until interrupted
158
        while ws_thread.is_alive():
159
            time.sleep(0.1)
160
    except KeyboardInterrupt:
161
        print("\nCtrl+C received. Stopping...")
162
        stop_event.set()  # Signal audio thread to stop
163
164
        # Send termination message to the server
165
        if ws_app and ws_app.sock and ws_app.sock.connected:
166
            try:
167
                terminate_message = {"type": "Terminate"}
168
                print(f"Sending termination message: {json.dumps(terminate_message)}")
169
                ws_app.send(json.dumps(terminate_message))
170
                # Give a moment for messages to process before forceful close
171
                time.sleep(5)
172
            except Exception as e:
173
                print(f"Error sending termination message: {e}")
174
175
        # Close the WebSocket connection (will trigger on_close)
176
        if ws_app:
177
            ws_app.close()
178
179
        # Wait for WebSocket thread to finish
180
        ws_thread.join(timeout=2.0)
181
182
    except Exception as e:
183
        print(f"\nAn unexpected error occurred: {e}")
184
        stop_event.set()
185
        if ws_app:
186
            ws_app.close()
187
        ws_thread.join(timeout=2.0)
188
189
    finally:
190
        # Final cleanup (already handled in on_close, but good as a fallback)
191
        if stream and stream.is_active():
192
            stream.stop_stream()
193
        if stream:
194
            stream.close()
195
        if audio:
196
            audio.terminate()
197
        print("Cleanup complete. Exiting.")
198
199
200
if __name__ == "__main__":
201
    run()

1
# Replace or establish new set of keyterms
2
websocket.send('{"type": "UpdateConfiguration", "keyterms_prompt": ["Universal-3"]}')
3
4
# Remove keyterms and reset context biasing
5
websocket.send('{"type": "UpdateConfiguration", "keyterms_prompt": []}')