Free, private, local dictation for Windows. Hold a hotkey, speak, and accurate text is typed wherever your cursor is — 100% on your device. The open-source voice-to-text alternative to Wispr Flow.
A free, open-source voice-to-text app that keeps your microphone on your own machine — the private alternative to paid cloud dictation.
Whisper runs locally on your CPU. Audio never leaves your PC — no cloud, no telemetry.
Terminals, Office, browsers, chat, IDEs — even elevated windows. Works where paste is blocked.
Auto-detected per phrase, with Hindi/Urdu correction. Dictate mixed-language meetings.
Toggle mode types as you speak — perfect for hands-free meeting notes.
A voice-activity gate ignores static and background noise instead of inventing text.
No GPU required. No fees, ever. MIT-licensed, open-source Python.
Grab Sotto-windows-x64.zip from Releases and unzip it.
SmartScreen may warn (unsigned) → More info → Run anyway.
Settings → Model → Download (one-time, ~300 MB).
Hold Ctrl + Win, talk, release. Text appears at your cursor. 🎉
| Sotto | Wispr Flow | Voice Access | Dragon | |
|---|---|---|---|---|
| Runs 100% offline | ✅ | ❌ | ✅ | ✅ |
| Free | ✅ | 💲 | ✅ | 💲💲 |
| Open source | ✅ | ❌ | ❌ | ❌ |
| Hindi / Hinglish | ✅ | ✅ | limited | ❌ |
| Type as you speak | ✅ | ✅ | ⚠️ | ✅ |
| No account needed | ✅ | ❌ | ✅ | ✅ |
The live dictation you're downloading is free and open source, forever. We're building an optional premium tier: upload any audio or video — any length, any language — and get an accurate transcript back. Hours-long recordings, meetings, lectures, podcasts, interviews. No limits.
Join the waitlist and we'll build it if enough people want it. No spam, no commitment — just tell us you're interested.
Practical, in-depth guides on free, offline voice-to-text for Windows.
Free, offline, private dictation apps for Windows compared honestly.
Set up free, 100% local dictation — no cloud, no subscription, no account.
Dictate mixed English-Hindi speech into any app, fully offline.
No. Transcription runs entirely on your CPU. The only network request is the one-time model download you start yourself.
Yes — MIT-licensed and free forever. The core voice-to-text dictation is open source. The upcoming unlimited full-file transcription is an optional paid tier that funds development.
No — it's tuned for CPU and runs on a normal laptop. You can optionally use an Intel iGPU.
The app is unsigned (certificates cost money; Sotto is free). Click More info → Run anyway, or build it from source.
English and Hindi/Hinglish out of the box, with one-click switching. Set "Auto — all languages" for ~99 Whisper languages.