Privacy · August 14, 2026

Your voice never leaves your Mac, literally

Every dictation product makes some version of the same promise: we respect your privacy. For almost all of them, it means your audio goes to their servers, gets transcribed there, and is handled under a promise to treat it responsibly.

We wanted to make a claim that is a property of the system rather than a policy about it, so Privt Voice runs the entire speech model on your Mac. When you hold Right Shift and talk, audio moves from your microphone into a transcription model executing on the Neural Engine, and the resulting text lands in whatever app you are using. The audio is then discarded from memory. No server takes part at any stage, because there is no server.

This approach became practical only recently. Modern open speech models have brought fast, accurate recognition down to a size that runs comfortably on Apple silicon, and CoreML executes them on the Neural Engine faster than a round-trip to any cloud API. On-device processing is no longer the compromise option: it is simply better, and it happens to be private.

Meetings too

The same engine transcribes calls, with your microphone on one lane and the call app's audio on the other, labelled Me and Them. Capture is scoped to the meeting app alone, so your music never enters a transcript, and the finished note is encrypted into your vault straight from memory, never written to disk in the clear.

Free forever is a consequence rather than a promotion. Software with no servers carries no marginal cost, so the local product will always cost nothing.