You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 

109 lines
3.9 KiB

# !/usr/bin/env python3
# -*- coding: utf-8 -*-
#
# Copyright 2016-2099 Ailemon.net
#
# This file is part of ASRT Speech Recognition Tool.
#
# ASRT is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# ASRT is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with ASRT. If not, see <https://www.gnu.org/licenses/>.
# ============================================================================
"""
@author: nl8590687
一个配置为可用于ASRT语音识别系统的录音程序
"""
import wave
import pyaudio
def record_wave(wavfile,
duration=10,
channels=1,
sampling_rate=16000,
sampling_bits=16,
chunk_size=1024,
keyboard_interrupt='keep_audio'):
"""Record audio using the default audio device by PyAudio and Wave"""
format_ = None
if sampling_bits == 8:
format_ = pyaudio.paInt8
if sampling_bits == 16:
format_ = pyaudio.paInt16
elif sampling_bits == 24:
format_ = pyaudio.paInt24
elif sampling_bits == 32:
format_ = pyaudio.paFloat32
else:
raise ValueError('Unsupported sampling bits')
audio = pyaudio.PyAudio()
stream = audio.open(format=format_,
channels=channels,
rate=sampling_rate,
input=True,
frames_per_buffer=chunk_size)
frames = []
print('Start to record with {}-seconds audio\n'
'Type Ctrl-C to get an early stop (a shorter audio)'
.format(duration))
try:
for _ in range(0, int(sampling_rate / chunk_size * duration)):
data = stream.read(chunk_size)
frames.append(data)
print('.', end='', flush=True)
except KeyboardInterrupt:
if keyboard_interrupt == 'keep_audio':
used_seconds = int(len(frames) * chunk_size / sampling_rate)
print('\n-*- Early stop with {} seconds'.format(used_seconds))
else:
raise
print('\nRecording finished')
stream.stop_stream()
stream.close()
audio.terminate()
print('Convert PCM frames to WAV... ', end='')
wavfp = wave.open(wavfile, 'wb')
wavfp.setnchannels(channels)
wavfp.setsampwidth(audio.get_sample_size(format_))
wavfp.setframerate(sampling_rate)
wavfp.writeframes(b''.join(frames))
wavfp.close()
print('OK')
if __name__ == "__main__":
from argparse import ArgumentParser, ArgumentDefaultsHelpFormatter
parser = ArgumentParser(description='Simple Wave Audio Recorder',
formatter_class=ArgumentDefaultsHelpFormatter)
parser.add_argument('-d', '--duration', type=int,
default=10, help='maximum duration in seconds')
parser.add_argument('-r', '--sampling-rate', type=int,
default=16000, help='sampling rate')
parser.add_argument('-b', '--sampling-bits', type=int,
default=16, choices=(8, 16, 24, 32), help='sampling bits')
parser.add_argument('-c', '--channels', type=int,
default=1, help='audio channels')
parser.add_argument('output', nargs='?', default='output.wav',
help='audio file to store audio stream')
args = parser.parse_args()
record_wave(args.output, duration=args.duration,
channels=args.channels,
sampling_bits=args.sampling_bits,
sampling_rate=args.sampling_rate)