Small script which ensures a somewhat consistent matrix live experience.
- Audio normalization to -16 LUFS
- Optional Whisper subtitle generation
- Whisper subtitle based show notes
- Python 3.12+
- ffmpeg
- ffmpeg-normalize
- whisper
pip install -r requirements.txt
python matrix_live_processor.py ./input.mp4 ./output.mp4 --use-whisper --model-name medium --language en
input_file
: The input file to processoutput_file
: The output file to write to (video filename)--use-whisper
: Enable whisper subtitle generation--burn-subtitles
: Burn the generated subtitles into the video--show-notes
: Generate show notes based on the generated subtitles. Requires ollama to be running.--model-name
: The whisper model to use (default:medium
). Check whispercpp for available models.--language
: The language to use for whisper subtitle generation (default:en
). Check whispercpp for available languages.--force
: Force overwrite of the output file if it already exists.
For best performance you should build the whisper lib manually. See https://github.com/abdeladim-s/pywhispercpp for details.
Apache 2.0