Three steps, no
setup ritual.
Hold Ctrl+Alt
Anywhere in Windows — a text field, a chat box, a code comment. No app to switch to first.
Say what you mean
A small floating widget turns red while it's listening, so you always know it's live.
Let go
Whisper transcribes it in under a second and the text is pasted right where your cursor was.
Built to disappear
into your workflow
Private by default
Local mode runs OpenAI's Whisper model entirely on your CPU. Audio never touches the internet.
Cloud mode when you need it
Flip one toggle to route through OpenAI's transcription API for noisy rooms or tricky accents.
Works in any app
No plugins or integrations to install per-app — it pastes into whatever currently has focus.
Free & open source
The full source is on GitHub. Read it, change it, or build it yourself from a clone.
Get Voice Flow
One executable, nothing to install. Unzips nowhere — just run it.
Windows 10/11 (64-bit) · Microphone · ~150 MB
free space for the speech model