Dictation for Mac

Voice to text in every macOS app, with punctuation, an offline engine, and commands.

macOS has had dictation built in for years, and most people who try it stop within a week. It makes you speak your punctuation aloud, it gives up after a while, and it hands back a block of text you then have to go and tidy. JotMic is the version of that feature people assumed they were getting.

Hold a key — Right Option by default, because it types nothing on its own and so costs you no shortcut you were already using. Speak normally. Release. Punctuated, capitalised text appears at your cursor, in whatever app you were already in.

It removes; it does not rewrite

The cleanup pass takes out the “um”s and the false starts and puts in the commas, full stops and capitals. That is the entire edit. It is bound not to touch anything else, so what lands is what you said with the noise taken out — not an AI’s improved version of your point. If you want even that off, one switch in Settings gives you the raw transcript.

Two engines, and the choice is yours per session

The hosted engine is faster on hard audio — accents, background noise, an indifferent microphone. The local engine runs a Parakeet model on your own Mac and makes no network request at all, which matters for anything under NDA and on any flight. Both are included on every plan, including the free trial. See offline dictation for what that actually guarantees.

The part macOS makes uniquely possible

On a Mac, JotMic can read the menu bar of whatever application is in front of you through the accessibility API — so every action that app advertises is already a spoken command, with no list to configure and nothing that goes stale when the app updates. Say “new folder” in Finder and a folder appears. Add to that opening applications, switching between them, tabs, and the system keys for volume, brightness and media.

This is the one thing on this page that no other dictation app for Mac does, and it is described in full on voice commands.

Setting it up

Two permissions and about two minutes. Microphone so it can hear you; Accessibility so it can type for you and read the menu bar. Nothing else is requested, and the app ships no screen-recording or camera entitlement whatsoever — the privacy page sets out exactly what is sent and what is stored.

Also on Windows

The same account covers both. Dictation, punctuation, language detection, custom vocabulary and modes all work on Windows 11; the menu-bar command layer is macOS-only for now.

Questions

Is this better than the dictation built into macOS?

It is a different thing in three ways. It punctuates and capitalises as you go instead of making you speak the punctuation, it does not stop after a fixed interval, and it can act on what you say — opening apps and firing menu items — rather than only typing it.

Which version of macOS do I need?

macOS 13 or later.

What permissions does it ask for, and why?

Two. Microphone, so it can hear you. Accessibility, so it can type at your cursor and read the frontmost app’s menu bar. It asks for nothing else — no screen recording and no camera entitlement are shipped at all, so it cannot capture your screen rather than merely choosing not to.

Does it work in every Mac app?

Any app or website with a text field. It inserts at the cursor the way a keyboard does, so there is nothing to integrate and no extension to install — Mail, Slack, Word, Pages, Xcode, Cursor, the terminal and everything else.

Can I use it on a Mac with no internet?

Yes. The local engine runs a Parakeet model on your own machine and needs no network at any point.

Sixty minutes free on both engines, no card. Long enough to find out whether you go back to the keyboard.

60 minutes free · no card