Pace is a lightning-fast local speech-to-text desktop application built for seamless dictation anywhere on your system.
Powered by Whisper (via faster-whisper) and built with Electron + Python, Pace provides instant transcription directly into your active window.
No cloud. No subscriptions. No latency.
Just speak Pace types.
Most dictation tools today rely on:
- cloud processing
- subscriptions
- closed ecosystems
Pace was built with a different philosophy.
✔ Local-first ✔ Open-source ✔ Community-driven ✔ Instant transcription
Your voice never leaves your machine.
Hold the hotkey, speak naturally, and release.
Your words appear exactly where your cursor is.
No switching apps. No copy-paste.
Pace uses faster-whisper to run speech recognition locally with extremely low latency.
No network delays. No waiting for servers.
Pace is fundamentally incapable of storing past audio.
Each session is:
- recorded
- transcribed
- immediately discarded
No logs. No recordings. No hidden storage.
Pace uses the Windows GetAsyncKeyState API for ultra-responsive hotkey detection.
This allows:
- instant recording start
- instant stop
- zero UI lag
Pace types directly into your active window.
It bypasses the clipboard entirely, eliminating:
- clipboard pollution
- delayed pastes
- “ghost” clipboard overwrites
A lightweight floating pill interface stays quietly on your screen.
Click to record. Right-click for options. Stay in your workflow.
All transcription happens locally.
No data collection. No telemetry. No external servers.
Pace automatically detects and cleans orphaned background processes to ensure stability.
No lingering background tasks.
Modern voice-dictation tools are powerful — but many come with trade-offs:
- subscription pricing
- cloud processing
- closed ecosystems
- limited transparency
Tools like Wispr Flow have demonstrated how powerful real-time dictation can be.
Pace takes that idea further.
Instead of a paid cloud product, Pace is designed as a local, open alternative that anyone can use, improve, and extend.
| Feature | Pace | Paid Dictation Apps |
|---|---|---|
| Price | Free & open source | Subscription |
| Speech Processing | Fully local | Often cloud |
| Privacy | Your data never leaves your machine | Depends on provider |
| Customization | Open source + community contributions | Limited |
| Latency | Near-instant local processing | Network dependent |
| Extensibility | Hackable | Closed ecosystem |
Pace is built on three core principles.
No servers. No telemetry. No hidden data collection.
You shouldn't need a monthly subscription to use your own voice.
Pace is designed to evolve through:
- contributions
- plugins
- community improvements
Pace isn't just another dictation tool.
The goal is to build the best open-source speech-to-text workflow for developers, writers, and creators.
Future plans include:
- plugin architecture
- smarter formatting commands
- multi-language dictation
- AI-assisted editing
- community-driven improvements
- Node.js (v18 or higher recommended)
- Python 3.11+
- FFmpeg (required by faster-whisper)
Clone the repository
git clone https://github.com/rofuniki-coder/Pace.git
cd Pace
Install Node dependencies
npm install
Install Python dependencies
pip install -r requirements.txt
Run the application
npm start
Hold:
Ctrl + Alt
Release to stop recording and instantly insert the transcription.
Click the floating Pace pill to start or stop recording.
Right-click the pill to:
- view transcription history
- hide the app for an hour
- check for updates
- Electron — Desktop interface
- Python — Speech processing backend
- Whisper (faster-whisper) — Local speech recognition
- FFmpeg — Audio processing
Contributions are welcome.
Ways to contribute:
- fix bugs
- suggest features
- improve documentation
- submit pull requests
Steps:
- Fork the repository
- Create a new branch
- Commit your changes
- Push to your branch
- Open a Pull Request
If you find Pace useful:
⭐ Star the repository 🐛 Report issues 💡 Suggest features 🔧 Contribute improvements
Every contribution helps make Pace the best open-source dictation tool.
This project is licensed under the MIT License.
Built by rofuniki-coder
Wispr Flow is a trademark of its respective owners. Pace is an independent open-source project and is not affiliated with or endorsed by Wispr Flow.
References to Wispr Flow are made strictly for comparison purposes.