Dictation for Windows
Voice to text in every Windows 11 app, with punctuation and an offline engine.
Windows Voice Typing works, in the sense that words appear. Then you go back through and add the commas, fix the capitals, and retype the three product names it has never heard of. The typing you saved comes back as editing.
JotMic is the same idea finished. Hold one key, speak normally, release. Punctuated, capitalised text lands at your cursor in whatever app you were already in — Outlook, Word, Teams, Slack, Chrome, VS Code, the terminal, anything with a text field.
It removes; it does not rewrite
The cleanup takes out fillers and false starts and puts in punctuation and capitals. That is the whole edit, and it is bound not to touch anything else — so what lands is what you said with the noise removed, rather than a rewritten version of your point. Turn it off in Settings and you get the raw transcript instead.
It learns the words you actually use
Names, acronyms, libraries and colleagues are where every dictation tool falls down. JotMic re-checks your custom vocabulary against what the microphone actually heard rather than fuzzy-matching what the recogniser guessed, which is how “post grist” becomes PostgREST instead of staying wrong in a slightly different way.
An engine that needs no network
Switch to the local engine and the whole pipeline runs on your own PC — Parakeet through sherpa-onnx, about 652 MB fetched once and then used from disk. No audio is uploaded and none is queued to be. The hosted engine is faster on difficult audio and does need a connection; offline dictation sets out the trade in full.
What is not here yet
On macOS, JotMic can also act on what you say — reading the frontmost application’s menu bar so that every menu item is a spoken command. That layer uses a macOS accessibility API and has no Windows equivalent today, so on Windows JotMic types and does not act. It is stated here rather than left to be discovered after you install it.
One account, both machines
If you move between a PC and a Mac, the same licence covers both and your vocabulary and modes follow you. JotMic on macOS describes what is different there.
Questions
Which version of Windows do I need?
Windows 11 on x64.
Is this better than Windows Voice Typing?
The differences people notice are punctuation that arrives correct without being dictated aloud, no cut-off after a pause, custom vocabulary that learns names and technical terms, and an offline engine you can switch to when nothing may leave the machine.
Does it work in every Windows app?
Any app or website with a text field — Outlook, Word, Teams, Slack, Chrome, VS Code, the terminal. It inserts at the cursor like a keyboard, so there is nothing to integrate.
Are voice commands available on Windows?
Not yet. Dictation, punctuation, language detection, custom vocabulary and modes all work. The command layer that reads an application’s menu bar uses a macOS API and has no Windows equivalent today.
Can I dictate on Windows with no internet?
Yes. The local engine runs Parakeet through sherpa-onnx on your own PC. The weights are about 652 MB, downloaded once, and after that nothing is fetched to transcribe.
Sixty minutes free on both engines, no card, on Windows 11.
60 minutes free · no card