helper/speech.py

41 lines
1.4 KiB
Python

import json
from urllib.request import Request, urlopen
from scipy.io import wavfile
class Speech:
@staticmethod
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}')
@staticmethod
def __load_wav(wav_file):
Speech.__check_wav(wav_file)
with open(wav_file, 'rb') as file:
result = file.read()
return result
@staticmethod
def __stt(wav_file: str, server: str) -> str:
print(f'Connecting to \'{server}\'...')
request = Request(url=f'{server}/stt',
data=Speech.__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}'
def run(self, wav_file, server):
return self.__stt(wav_file, server)