Transcriber

A utility which uses the whisper.cpp to transcribe audio files. I wrote this literally hours before @charliemarsh released this, but it was a fun task to use Pixi to build things from source.

Getting started

Install Pixi curl -fsSL https://pixi.sh/install.sh | bash
Set up the whisper model: pixi run install_whisper
- Downloads and unzips the whisper.cpp repo
- Builds the from source
- Downloads tiny.en model
- Cleans up build
Run example: python -m transcriber | jq -c (Will transcribe "samples/bruce.mp3")
- Transcriber class will auto-convert audio files to WAV format at runtime

streamlit run app.py