helper/main.py

#!/usr/bin/env python3

import json
import sys
from urllib.request import Request, urlopen

from scipy.io import wavfile


def stt(wav_file: str, url: str) -> str:
    print(f'Connecting to \'{url}\'...')
    request = Request(f'{url}/stt', data=_load_wav(wav_file), headers={'Content-Type': 'audio/wav'})
    result = json.loads(urlopen(request).read().decode('utf-8'))

    if not ('code' in result and 'text' in result):
        raise RuntimeError(f'Wrong reply from server: {result}')
    return result.text if not result.code else f'Server error: [{result.text}]: {result.code}'


def _load_wav(wav_file):
    _check_wav(wav_file)
    with open(wav_file, 'rb') as file:
        result = file.read()
    return result


def _check_wav(wav_file):
    sample_rate, sig = wavfile.read(wav_file)
    channels = len(sig.shape)
    bits = sig.dtype.base.name
    if sample_rate != 16000:
        raise Exception(f'Sample rate is not 16000: {sample_rate}')
    if channels != 1:
        raise Exception(f'Number of Channels is not 1 (Not mono): {channels}')
    if bits != 'int16':
        raise Exception(f'Bits per sample 16: {bits}')


def _main():
    if len(sys.argv) < 2:
        print('Usage: {} FILE [URL]'.format(sys.argv[0]))
        exit(1)
    file = sys.argv[1]
    server = 'http://127.0.0.1:8086' if len(sys.argv) < 3 else sys.argv[2]
    print(f'Result: {stt(file, server)}')


if __name__ == '__main__':
    _main()