Medical Mode

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
    "domain": "medical-v1"
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
                break
54
        print("Audio streaming stopped.")
55
56
    global audio_thread
57
    audio_thread = threading.Thread(target=stream_audio)
58
    audio_thread.daemon = True
59
    audio_thread.start()
60
61
def on_message(ws, message):
62
    try:
63
        data = json.loads(message)
64
        msg_type = data.get('type')
65
66
        if msg_type == "Begin":
67
            session_id = data.get('id')
68
            expires_at = data.get('expires_at')
69
            print(f"\nSession began: ID={session_id}, ExpiresAt={datetime.fromtimestamp(expires_at)}")
70
        elif msg_type == "Turn":
71
            transcript = data.get('transcript', '')
72
            formatted = data.get('turn_is_formatted', False)
73
74
            if formatted:
75
                print('\r' + ' ' * 80 + '\r', end='')
76
                print(transcript)
77
            else:
78
                print(f"\r{transcript}", end='')
79
        elif msg_type == "Termination":
80
            audio_duration = data.get('audio_duration_seconds', 0)
81
            session_duration = data.get('session_duration_seconds', 0)
82
            print(f"\nSession Terminated: Audio Duration={audio_duration}s, Session Duration={session_duration}s")
83
    except json.JSONDecodeError as e:
84
        print(f"Error decoding message: {e}")
85
    except Exception as e:
86
        print(f"Error handling message: {e}")
87
88
def on_error(ws, error):
89
    """Called when a WebSocket error occurs."""
90
    print(f"\nWebSocket Error: {error}")
91
    stop_event.set()
92
93
94
def on_close(ws, close_status_code, close_msg):
95
    """Called when the WebSocket connection is closed."""
96
    print(f"\nWebSocket Disconnected: Status={close_status_code}, Msg={close_msg}")
97
98
    global stream, audio
99
    stop_event.set()
100
101
    if stream:
102
        if stream.is_active():
103
            stream.stop_stream()
104
        stream.close()
105
        stream = None
106
    if audio:
107
        audio.terminate()
108
        audio = None
109
    if audio_thread and audio_thread.is_alive():
110
        audio_thread.join(timeout=1.0)
111
112
113
# --- Main Execution ---
114
def run():
115
    global audio, stream, ws_app
116
117
    audio = pyaudio.PyAudio()
118
119
    try:
120
        stream = audio.open(
121
            input=True,
122
            frames_per_buffer=FRAMES_PER_BUFFER,
123
            channels=CHANNELS,
124
            format=FORMAT,
125
            rate=SAMPLE_RATE,
126
        )
127
        print("Microphone stream opened successfully.")
128
        print("Speak into your microphone. Press Ctrl+C to stop.")
129
    except Exception as e:
130
        print(f"Error opening microphone stream: {e}")
131
        if audio:
132
            audio.terminate()
133
        return
134
135
    ws_app = websocket.WebSocketApp(
136
        API_ENDPOINT,
137
        header={"Authorization": YOUR_API_KEY},
138
        on_open=on_open,
139
        on_message=on_message,
140
        on_error=on_error,
141
        on_close=on_close,
142
    )
143
144
    ws_thread = threading.Thread(target=ws_app.run_forever)
145
    ws_thread.daemon = True
146
    ws_thread.start()
147
148
    try:
149
        while ws_thread.is_alive():
150
            time.sleep(0.1)
151
    except KeyboardInterrupt:
152
        print("\nCtrl+C received. Stopping...")
153
        stop_event.set()
154
155
        if ws_app and ws_app.sock and ws_app.sock.connected:
156
            try:
157
                terminate_message = {"type": "Terminate"}
158
                print(f"Sending termination message: {json.dumps(terminate_message)}")
159
                ws_app.send(json.dumps(terminate_message))
160
                time.sleep(5)
161
            except Exception as e:
162
                print(f"Error sending termination message: {e}")
163
164
        if ws_app:
165
            ws_app.close()
166
167
        ws_thread.join(timeout=2.0)
168
169
    except Exception as e:
170
        print(f"\nAn unexpected error occurred: {e}")
171
        stop_event.set()
172
        if ws_app:
173
            ws_app.close()
174
        ws_thread.join(timeout=2.0)
175
176
    finally:
177
        if stream and stream.is_active():
178
            stream.stop_stream()
179
        if stream:
180
            stream.close()
181
        if audio:
182
            audio.terminate()
183
        print("Cleanup complete. Exiting.")
184
185
186
if __name__ == "__main__":
187
    run()

1
CONNECTION_PARAMS = {
2
    "sample_rate": 16000,
3
    "speech_model": "u3-rt-pro",
4
    "domain": "medical-v1",
5
    "speaker_labels": "true",
6
    "keyterms_prompt": json.dumps(["Lisinopril", "Metformin", "Humalog"])
7
}

1
CONNECTION_PARAMS = {
2
    "sample_rate": 16000,
3
    "speech_model": "u3-rt-pro",
4
    "domain": "medical-v1",
5
    "min_turn_silence": 800,
6
    "max_turn_silence": 3600,
7
}