【ASR】wav文件位数转换

tensorflow.python.framework.errors_impl.InvalidArgumentError: Can only read 16-bit WAV files, but received 24

I would suggest using SoX for this task. Changing the bit depth is very simple:

sox old.wav -b 16 new.wav

If you must use Python, then you could use PySoundFile as you found. Here’s a little code snippet:

import soundfile

data, samplerate = soundfile.read('old.wav')
soundfile.write('new.wav', data, samplerate, subtype='PCM_16')

You should also use soundfile.available_subtypes to see which subtypes you can convert a file to. Here’s its sample usage, taken from their documentation:

>>> import soundfile as sf
>>> sf.available_subtypes('FLAC')
{'PCM_24': 'Signed 24 bit PCM',
 'PCM_16': 'Signed 16 bit PCM',
 'PCM_S8': 'Signed 8 bit PCM'}

How to convert a 24-bit wav file to 16 or 32 bit files in python3