For macOS · Apple silicon

Briska for macOS.

macOS 14.5 or later · Apple silicon · English-only transcription

What you download

Downloaded once. Then it's yours.

Cloud dictation downloads almost nothing, because the model lives on someone else's computer, and so does your voice. Briska brings the models to your Mac instead: three pieces to get started, two more only if you want them, and after that nothing needs the network.

BriskaThe app, and the engine that runs the modelsOne-time
Speech modelTurns your voice into words. English onlyOne-time
Language modelTurns those words into finished writingOne-time
Voice commandsTurns a spoken instruction into one actionSettings
Meeting speaker labelsWorks out who spoke on the other side of a callSettings

About 3.2 GB in total for the three you start with, verified against a fingerprint as it arrives, and resumable if your connection drops midway. The last two are fetched only if you turn those features on in Settings; they are a few hundred megabytes between them, and you may never want either.

What it needs

Requirements, and the honest limits.

Ready when you are.

An address, not an account, then the download. Nothing is sent now, and what does arrive is rare: a new version, or a change worth hearing from us first.