hold, or go hands-free
Double-tap the shortcut and keep talking for up to five minutes. Tap again to finish, Esc to throw it away.
Hold your shortcut, say it, let go. The text appears at your cursor in under half a second, and the speech recognition runs on your own computer.
Hey Leo,
The login page is throwing errors again since this morning's deploy. Can you roll it back while I check the logs?
let go → inserted in 0.31 sBurbuja sits in your tray and wakes up when you press the shortcut, in any app.
Press your shortcut in any app. You choose the keys, and the mic opens in about 30 ms.
Talk the way you'd talk to a coworker, pauses included. Whisper understands English, Spanish and 97 other languages.
Release the keys and the text appears at your cursor, punctuated and ready to send.
Cloud dictation sends your voice to someone else's server. Burbuja has no server. The speech model runs on your computer, and the only thing the app downloads is the model you pick.
You never create an account, and the app sends no usage data. When you delete your history, it's erased from the disk.
Double-tap the shortcut and keep talking for up to five minutes. Tap again to finish, Esc to throw it away.
post gres → PostgreSQL
Connect a local model through Ollama or LM Studio to remove filler words and fix punctuation. It adds about 0.2 s.
Search, copy or re-polish past dictations, and choose how many days to keep them.
The app speaks English and Spanish. Whisper understands 99 languages and figures out which one you're using.
Install it, pick a speech model and hold your shortcut.
Free and open source under the MIT license.