I have been working on a customizable AI voice dictation tool using Pipecat's framework to swap between many providers and models, including cloud or local.

Started off as an open source alternative to Wispr Flow for myself as I wanted to have more control over the formatting rules as well as model choice but after sharing with friends and presenting it at my local Claude Code meetup, I was encouraged to share it more widely.

The desktop app uses tauri so it is cross-platform compatible and I have tested it working on macOS and windows.

https://github.com/kstonekuan/tambourine-voice