$ pip install git+https://github.com/openai/whisper.git
$ ffmpeg -ss 00:00:00 -to 00:10:00 -i ~/Downloads/KJ法勉強会振り返り勉強会.mp4 -c copy tmp.mp4
$ whisper tmp.mp4 --language Japanese --model large
-
UserWarning: FP16 is not supported on CPU; using FP32 instead -
whisper tmp.mp4 --language Japanese --model large 617.96s user 340.21s system 188% cpu 8:28.80 total -
--device mps -
[01:03:27.660 --> 01:03:55.660] ご清聴ありがとうございました- whisper tmp.mp4 —language Japanese —model large 38903.09s user 20685.49s system 150% cpu 11:01:14.63 total
https://fukugyouhistory.tokyo/?p=1207 GitHub - guillaumekln/faster-whisper: Faster Whisper transcription with CTranslate2