Voice-to-text that runs entirely on your machine. Private, fast, and free. No cloud, no account, no subscriptions.
100% free. Open source. Your audio never leaves your computer.
Hold a hotkey, say what you need, and the text appears wherever your cursor is. That's it.
Hold your hotkey and speak. Release to transcribe. Text gets pasted right where your cursor is, in any app.
All transcription happens on your machine. No audio is sent anywhere. Your conversations stay yours.
From tiny (instant results) to large-v3 (max accuracy), plus Distil-Whisper for the best speed-to-quality ratio.
Auto-detects your NVIDIA GPU for fast transcription. No GPU? Falls back to CPU automatically.
Every transcription is saved. Copy, delete, or scroll back through everything you've dictated.
Set your own push-to-talk key combo. Toggle hands-free mode when you want to dictate without holding anything.
No installer. No account. No setup wizard.
Grab VoiceFlow.zip from GitHub. Extract it anywhere.
Double-click VoiceFlow.exe. Pick a Whisper model on first launch.
Hold Ctrl + Win and talk. Release and the text appears at your cursor.
That's it. Works in any app. Email, Slack, your code editor, anywhere.
Same idea, different approach. VoiceFlow keeps everything local and costs nothing.
| VoiceFlow | Wispr Flow | |
|---|---|---|
| Price | Free forever | $8-12/month |
| Privacy | ✓ 100% local | ✗ Cloud-based |
| Open source | ✓ Yes | ✗ Closed source |
| Account required | ✓ None | ✗ Required |
| Model options | 13 models | 1 (their API) |
| GPU acceleration | ✓ CUDA + CPU fallback | N/A (cloud) |
| Works offline | ✓ Yes | ✗ No |
13 Whisper models available out of the box. Downloaded on first use, stored locally.