docs
Get started
Everything needed to get from a download to a working shortcut. Nothing here requires an account, and none of it needs a network connection after the model is on disk.
Updated · Vitalii Yemets
Three steps to the first dictation
Download and install
Two builds, same program. Pick whichever suits the machine. The installer can fetch a speech model on the way; the portable archive runs from any folder.
Install →First launch
The first run opens a wizard of five short steps. Each can be skipped, and none of them asks for anything beyond what the program needs to work.
First launch →Hold, speak, let go
Put the caret where text can be typed, hold ctrl + win, say a sentence, let go. The text lands at the caret.
Your first dictation →Which model?
The wizard picks a sensible one. If your language has a narrow model of its own it is faster and usually more accurate than the general one, and the catalogue says which is which.
Choose a speech model for your language →
Everything in the docs
Installer and portable build, silent install, the SmartScreen warning.
The five-step wizard, microphone permission, your first dictation.
Which model for which need, and what each one in the catalogue is good at.
Every key combination and how to change them.
One row per language: which model recognises it, and translation.
Punctuation, the dictionary of rare words, replacements, voice commands.
A second model that rewrites the text to your prompt, local or external.
Keeping dictations, for how long, and which programs to leave out.
Every config.json key, every command-line flag, files and ports.
One script and Docker Desktop; what ends up in dist.
The five things that go wrong most often, and the questions people ask.
Wondering whether you need this at all? How to dictate offline on Windows 11 walks through Win+H, Voice Access and local models. Working in an editor? VS Code, the terminal and SSH. Typing hurts? Typing by voice without the mouse.