mic-system/stt_worker.py

#!/usr/bin/env python3
"""STT worker subprocess — connects to stt.mm.mk via WebSocket.

Receives PCM via UDP, sends to STT server.
Posts JSON results back to Flask via HTTP POST.

Usage: python3 stt_worker.py <ws_url> --audio-port 19876 --callback-url http://127.0.0.1:5000/internal/stt
"""
import argparse
import json
import socket
import threading
import urllib.request

import websocket


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("url")
    parser.add_argument("--audio-port", type=int, default=19876)
    parser.add_argument("--callback-url", default="http://127.0.0.1:5000/internal/stt")
    args = parser.parse_args()

    connected = False
    ws = None

    def emit(obj):
        try:
            data = json.dumps(obj).encode("utf-8")
            req = urllib.request.Request(
                args.callback_url,
                data=data,
                headers={"Content-Type": "application/json"},
                method="POST",
            )
            urllib.request.urlopen(req, timeout=2)
        except Exception:
            pass

    def on_open(w):
        nonlocal connected
        connected = True
        emit({"type": "stt_status", "connected": True})

    def on_message(w, message):
        try:
            emit(json.loads(message))
        except Exception:
            pass

    def on_error(w, error):
        pass

    def on_close(w, code, msg):
        nonlocal connected
        connected = False
        emit({"type": "stt_status", "connected": False})

    ws = websocket.WebSocketApp(
        args.url,
        on_open=on_open,
        on_message=on_message,
        on_error=on_error,
        on_close=on_close,
    )

    ws_thread = threading.Thread(
        target=ws.run_forever,
        kwargs={"ping_interval": 20, "ping_timeout": 10},
        daemon=True,
    )
    ws_thread.start()

    # Listen for audio on UDP
    audio_sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
    audio_sock.bind(("127.0.0.1", args.audio_port))
    audio_sock.settimeout(1.0)

    try:
        while True:
            try:
                data, _ = audio_sock.recvfrom(65536)
            except socket.timeout:
                continue
            except OSError:
                break
            if connected and ws and data:
                try:
                    ws.send(data, opcode=0x2)
                except Exception:
                    pass
    except KeyboardInterrupt:
        pass
    finally:
        if ws:
            ws.close()
        audio_sock.close()


if __name__ == "__main__":
    main()