I built this because I wanted to see how far I could get with a voice-to-text app that used 100% local models so no data left my computer. I've been using a ton for coding and emails. Experimenting with using it as a voice interface for my other agents too. 100% open-source MIT license, would love feedback, PRs, and ideas on where to take it.