Dictami was designed around the Neural Engine in every Apple Silicon chip. Recognition runs entirely on your Mac — no server, no upload, no delay from a network round trip. On an M1 it's already fast. On M2, M3, or M4 it's faster still.
Before Apple Silicon, running a speech recognition model locally meant waiting several seconds per sentence — too slow to be useful. The Neural Engine built into every M-series chip changed that. It can run the Whisper recognition model at a fraction of the time it takes on Intel, which is why local dictation now feels instant rather than sluggish.
Dictami uses a runtime compiled specifically for Apple Silicon. It doesn't fall back to a generic CPU path — the Neural Engine does the work, and the result is ready in about a third of a second for a typical sentence.
Cloud dictation services exist because running recognition locally used to be too slow. That's no longer true on Apple Silicon. Dictami keeps the model loaded in memory, so when you release the key the audio is processed immediately on the chip. Nothing is uploaded. It works without an internet connection.
Hold the dictation key, speak, release. The text appears wherever your cursor was — a browser tab, an email, a notes app, a code editor. Automatic punctuation and capitalization mean the output is ready to send without editing.
Every supported language runs on the same chip at the same speed: English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Russian, Japanese, Korean, Chinese, and 16 more. No switching models. No extra downloads per language.
Dictami requires macOS 14 Sonoma or later and any Apple Silicon Mac. Intel Macs aren't supported — the runtime is built specifically for the Neural Engine and won't run on Intel. Any M-series chip works: MacBook Air, MacBook Pro, Mac mini, Mac Studio, or Mac Pro.
7-day trial inside the app. No card, no account.
Download Dictami