New Voice Engine · 3 Clicks 1.20
Voice Engine lets you pick the speech-to-text model behind every dictation. Stay on 3 Clicks Cloud for the best accuracy, or switch to Apple Speech to transcribe right on your Mac.
Workspace / Voice Engine
Choose the speech-to-text model dictation uses.
Active model
Streams your voice to Deepgram Nova-3 while you speak, with OpenAI as a fallback. The most accurate option, and it learns the words in your Dictionary.
Available models
Click a model to switch to it. Speed and accuracy are the app’s relative ratings.
Two engines, one shortcut
Your shortcut, transforms, history and AI cleanup work the same way on both. Only the transcription step changes.
Your voice streams to Deepgram Nova-3 as you speak, with OpenAI ready as a backup.
Best for: everyday dictation where every word matters.
Apple’s speech model turns your speech into text right on your Mac, with no cloud speech-to-text.
Best for: keeping transcription on your own Mac.
What happens when you speak
Both engines run AI cleanup in the cloud before anything is pasted, so your text still gets punctuation, filler-word removal and formatting. With Apple Speech, only the transcription itself stays on your Mac.
Press your shortcut and talk, the same as always.
Your audio is sent securely to our speech providers.
The text is polished in the cloud on both engines.
The finished text lands at your cursor.
Built not to fail you
If Apple Speech can’t finish a recording, that recording is transcribed in the cloud instead, so no words are lost.
While the Apple Speech model downloads, or on Macs older than macOS 26, dictation simply keeps using 3 Clicks Cloud.
The engine is chosen when you start recording, so switching mid-dictation never mixes the two.
Switch in seconds
In the 3 Clicks window, choose Voice Engine in the sidebar.
Click Apple Speech or 3 Clicks Cloud to preview its speed and accuracy.
Apple Speech downloads its model once. Your next dictation uses the new engine.
Questions
No. Apple Speech keeps the transcription step on your Mac. The resulting text is still sent to the cloud for AI cleanup before it’s pasted, and, as with every dictation, the recording is kept on our servers for up to 30 days for quality review, then deleted automatically. If Apple Speech can’t finish a recording, that recording is transcribed in the cloud instead. See the privacy policy.
3 Clicks Cloud is the default and the most accurate, and it learns the words in your Dictionary. Choose Apple Speech if you’d rather your speech is transcribed on your own Mac.
Apple Speech needs macOS 26 or later and Apple’s English speech model. On older macOS versions, or Macs where it isn’t supported, 3 Clicks keeps using 3 Clicks Cloud.
Both engines currently support English.
Voice Engine is in 3 Clicks 1.20 and later. Download it below, or let the app update itself.
Available now
Voice Engine is included in 3 Clicks 1.20 for macOS.