Dictate freely with local AI. Zero latency. Zero data leaks. Zero cost.
VoiceFlow brings OpenAI's Whisper directly to your Windows machine. Every word you speak is processed entirely on your hardware—your voice data never leaves your device. Built for privacy-conscious professionals who demand speed and reliability.
Cloud dictation services charge monthly fees while harvesting your voice data. VoiceFlow is free, fully local, and yours forever.
| Feature | VoiceFlow | Cloud Services |
|---|---|---|
| Cost | $0.00 | $10-15/mo |
| Data Privacy | 100% Local | Cloud Processed |
| Offline Support | Full Capability | None |
| Latency | Real-time | Network Dependent |
| Account Required | No | Yes |
| Open Source | MIT License | Proprietary |
Everything runs on localhost. Your microphone data never leaves your RAM. We can't sell your data because we never see it.
- Air-Gapped Safe: Works completely offline after initial model download.
- Open Source: Audit every line of code yourself.
- No Telemetry: Zero tracking, zero analytics, zero cloud calls.
No hidden processes, no cloud uploads. Just transparent, local AI at every step.
VoiceFlow waits silently in your system tray. A minimal popup indicates recording status.
Activate with your hotkey and speak naturally. Audio stays in RAM only—the interface visualizes your voice amplitude in real-time.
Release the hotkey. Local AI processes your audio instantly, then auto-pastes text at your cursor.
Configure your preferred keyboard shortcuts with two recording modes to match your workflow.
- Hold Mode: Hold to record, release to transcribe. Perfect for quick dictation bursts.
- Toggle Mode: Press once to start, press again to stop. Ideal for longer recordings.
Choose from 16+ Whisper models optimized for different use cases.
- Standard (Tiny → Large-v3): From 75MB to 3GB. Balance speed and accuracy for your hardware.
- Turbo (~1.6GB): Best speed-to-quality ratio. Recommended for daily use.
- English-only (.en variants): Optimized specifically for English with improved accuracy.
- Distilled: Faster inference with minimal quality loss.
- 99+ Languages: Automatic language detection built-in.
- Custom Hotkeys: Configure your own shortcuts with Hold or Toggle modes.
- Local History: Searchable SQLite database of all your transcriptions.
- Auto-Paste: Text appears directly at your cursor—no copy-paste needed.
Take back control of your voice data. Open source and forever free.
Windows 10/11 • 64-bit • ~150MB
Build and contribute to VoiceFlow.
# Clone and setup
git clone https://github.com/infiniV/VoiceFlow.git
cd VoiceFlow
pnpm run setup
# Development with hot-reload
pnpm run dev
# Build installer
pnpm run build:installer| Layer | Technology |
|---|---|
| Core | Pyloid (PySide6 + QtWebEngine) |
| Inference | faster-whisper (CTranslate2) |
| Frontend | React 18, Vite, Tailwind CSS v4 |
| UI | shadcn/ui, Lucide React |