This is a demo showing how to transcribe an .m4a audio file using OpenAI Whisper.
- Ubuntu 20.04 LTS Focal
- ffmpeg 4.2.7
- pytorch == 2.0.1
- transformers == 4.29.2
See requirements.txt
Please set REQUESTS_CA_BUNDLE according. E.g.
REQUESTS_CA_BUNDLE="/home/<user>/certs/cacert.pem"
The audio sample needs to be downsampled into 16kHz, and .WAV format. FFMPEG can be used as follows:
ffmpeg -i magnets.m4a -ar 16000 magnets16K.wav