I built a meeting recorder for Omarchy

This morning I asked Claude a simple question: can Voxtype, the voice-to-text app by Pete Jackson, record a meeting, including the audio of the other side? It can, but not as separate tracks, and you can’t edit the transcript or add chapters. So by the end of the day I had built Meeting Recorder, a meeting recorder for Omarchy. :-)

It records my microphone and whatever my computer plays as two separate tracks. No bot joins the call and it works with every meeting app, because it just listens to what I say and what my speakers play. When I stop, whisper.cpp transcribes it on my own machine. I’m on one track and the rest of the call is on the other, so it knows who said what without guessing.

Transcribing takes a little while, so I gave it a 90s animation: a neon sun over a scrolling grid, with the lines going by as they are recognised. Completely unnecessary, and it stays on screen for at least ten seconds, because otherwise you’d miss it.

When it’s done you get a player with a waveform. Click any line in the transcript and it plays from there. You can fix a line right where you read it, give it to the other speaker, or delete it. Chapters come from whatever coding agent Omarchy uses (Claude Code, in my case), run without any tools, so all it does is read the transcript.

You can also drop in a recording you already have. Then there is only one track, so it tells the voices apart by itself.

And of course it wears your Omarchy theme, and switches along when you change it.

It started as a Python script and was a Rust app with GTK and libadwaita by the afternoon. Install it with one line:

curl -fsSL https://raw.githubusercontent.com/jankeesvw/omarchy-meeting-recorder/main/install.sh | bash

That builds a proper pacman package, so pacman -R removes it again. I also opened a pull request to get it into the Omarchy package repository. The code is on GitHub, MIT licensed.

Hello, I'm Jankees van Woezik

Like this post? Follow me on X (@jankeesvw)