Dictation for macOS. Press a key, speak, press it again, and the words appear at your cursor.
What you say stays on your Mac. Recognition runs on curated open models, quantized to perform well on Apple silicon, with no provider between you and your words. A calibration pass reads your own voice against several models and picks the fit, measured rather than assumed.
It started as a personal problem: dictating code and prose all day without renting a subscription or sending audio to someone else's server.
Wavelet is in a friendly beta today, with a public release on its way. To try it, write to hello@littoralai.com.