100% offline — your audio never leaves your machine
Speak. It types.
Saut turns your voice into text in any app on your computer. Press a shortcut, say what you mean, and watch it appear.
Free forever · No account · Windows 10 or 11 (64-bit)
Hey Sarah — can you send over the Q3 numbers before the standup? Thanks!
Why voice
You already speak 3× faster than you type
Most people type around 40 words a minute and speak around 130. Saut closes that gap in every message, document, and reply you write.
New in 1.3
Saut has a face
Hold the shortcut and a small robot boots up at the bottom of your screen. It wears Saut's waveform on its chest, and the bars move with your voice — so you can always tell it's hearing you. Let go and it nods, thinks for a beat, and types.
-
Listening
The bars on its chest ride your voice, and the halo on its antenna means the mic is live. You never have to guess whether Saut heard you.
-
Thinking
Let go and it nods — got it — then the wave ripples while the model turns what you said into text, right there on your machine.
-
Out of your way
Drag it anywhere on screen, swap it for a quiet waveform pill, or switch the on-screen indicator off entirely. It's your desktop.
Shown large. On screen it's about 90 px tall.
How it works
Three keys to a sentence
There is no app to switch to. Saut lives in the background and types where you already are.
- 01
Press the shortcut
Hold the shortcut in any app — your editor, your inbox, a chat box. Saut starts listening instantly.
Ctrl + Space
- 02
Just speak
Talk normally. A little robot boots up at the bottom of your screen and its chest moves with your voice, so you can see Saut is hearing you — and everything stays on your machine.
- 03
Your words appear
Let go. The robot nods, thinks for a moment, and the text is typed straight into whatever you were using — no copy-paste, no switching windows.
Features
Built to stay out of your way
Everything you need for dictation that feels native — and nothing you didn't ask for.
Runs entirely offline
The speech model lives on your computer. Your voice never touches a server — not ours, not anyone's.
Works in every app
Saut types into whatever has focus. Slack, VS Code, Gmail, Word, a terminal — it doesn't care.
Push-to-talk or toggle
Hold the key while you speak, or tap once to start and once to stop. Both are one keystroke away.
Fifteen speech models
Whisper, Parakeet, Moonshine, Canary and more — pick the one that fits your machine, from a 31 MB featherweight to the most accurate.
Speaks your language
Whisper models handle 99 languages, and Saut's own interface ships in 19.
Free, with no account
No sign-up, no subscription, no trial that expires. Download it and it works.
A robot, or a quiet pill
The robot shows up while you dictate. Prefer something calmer? Switch to a compact waveform pill, drag either anywhere, or turn it off.
Transcripts stay on your disk
Saut keeps a local history you can search, copy from, or wipe. It is a file on your machine, not a row in someone's database.
Small and quiet
A 13 MB download that lives in your tray and unloads the speech model from memory when you stop using it.
Privacy
Your voice stays yours
Most dictation tools stream your microphone to a server. Saut doesn't. The model runs on your own hardware, so your audio — and whatever you happened to say — never leaves the machine you said it on.
Read how it works- No account and no sign-up
- No telemetry, no analytics, no crash reports
- Audio is processed 100% on your device
- Works with the internet switched off
FAQ
Questions, answered
Is Saut really free?
Yes. There's no trial, no paid tier, no usage cap, and no account. You download it, and it works — that's the whole model.
Does my voice get uploaded anywhere?
No. The speech model runs on your own computer, and Saut keeps working with the internet switched off — which is the simplest proof that nothing is being sent anywhere.
Which apps does it work in?
Any app you can type in. Saut types into whatever window has focus — email, documents, chat, code editors, browsers, terminals.
What languages does it support?
Whisper models transcribe 99 languages, and Saut's own interface is available in 19.
What do I need to run it?
Windows 10 or 11 (64-bit), a microphone, and room for a speech model — about 450 MB for the recommended one, up to 1.7 GB for the largest. The app itself is a 13 MB download.
Can I turn the robot off?
Yes. You can swap it for a compact waveform pill, drag either one anywhere on your screen, or switch the on-screen indicator off entirely — all in Settings. Saut works exactly the same either way.
Why does Windows show a warning when I install?
Saut isn't code-signed yet, and Windows flags any installer without a paid certificate. Click “More info”, then “Run anyway” — the app is safe, and you can verify every release on GitHub.
Stop typing what you could just say
Download Saut and try it in the next thing you write.
13 MB · Windows 10 or 11 (64-bit) · macOS and Linux coming soon