Hold a key. Talk. Get clean text.
Voice dictation for every app on Windows. Speech recognition runs on your own PC, so your voice is never uploaded.
People speak at roughly 150 words a minute and type at roughly 40. WhisperFlow closes that gap: hold a key, say what you mean, let go, and the finished text lands wherever your cursor is. Your editor, your terminal, your email, a chat window.
Every other dictation tool worth using ships your voice to someone else's server. WhisperFlow does not. Recognition runs locally, cleanup is a set of plain rules rather than a language model, and transcripts never sync.
Types into any app
Not a text box you copy from. It types into whichever window has focus, via a global hotkey.
Runs on your PC
Speech recognition is local (whisper.cpp and sherpa-onnx). After the engine downloads, dictation needs no connection at all.
Cleans up after you
Fillers go, punctuation appears. Deterministic rules, so it never rewrites what you meant.
Learns your words
A personal dictionary for names and jargon, and snippets that expand two words into a whole paragraph.
A closer look
Everything it does
Dictation
- Hold-to-talk or tap-to-toggle global hotkey
- Types into the focused window of any app
- "Scratch that" discards the last dictation
- Voice commands: bullet that, make it formal, make it an email
Recognition
- 99 languages via Whisper
- Faster European and CJK engines available
- Works fully offline once the engine is installed
- Translation to English while you speak
Your vocabulary
- Personal dictionary with "sounds like" hints
- Snippets that expand short phrases
- Settings, dictionary and snippets sync across your own PCs
- Transcripts are never synced
Free for 7 days, then $2/month
Billed through the Microsoft Store. Cancel any time.
- Platform
- Windows 10 and 11
- Category
- Productivity
- Publisher
- Technocrux, on the Microsoft Store
- Legal
- Privacy policy and terms of use
