Minimalist AI speech transcription for desktop linux

Find a file

ns 8fc9c9247d added text cleanup step after transcription		2026-06-04 23:27:25 -05:00
.gitignore	inital commit	2026-06-04 16:25:22 -05:00
.python-version	inital commit	2026-06-04 16:25:22 -05:00
auto.py	inital commit	2026-06-04 16:25:22 -05:00
LICENSE.txt	added license	2026-06-04 16:30:17 -05:00
main.py	added text cleanup step after transcription	2026-06-04 23:27:25 -05:00
pyproject.toml	inital commit	2026-06-04 16:25:22 -05:00
README.md	waybar integration	2026-06-04 18:20:05 -05:00
uv.lock	inital commit	2026-06-04 16:25:22 -05:00

README.md

simplewhispr

minimalist ai-powered dictation for linux.

requires ffmpeg and wtype.

install

uv tool install . --reinstall

usage

run simplewhispr-auto to start dictating. run simplewhispr-auto again when you are done. wait a few seconds, and what you said will be transcribed onto your screen with wtype.

requires ~/.config/simplewhispr/config.json:

{
  "openai_api_key": "sk-...",
  "model": "gpt-4o-mini-transcribe" // optional, defaults to mini
}

integrations

waybar: tail -F /tmp/simplewhispr-waybar.log, json output