Open source Typeless alternative — AI voice input with real-time polish.

Voix is a local-first menu bar application that transforms your speech into polished, professional text in any application. Speak naturally, and Voix handles the rest.

1. Press a hotkey (Option+E on macOS, Alt+E on Windows)
2. Speak — your words are transcribed locally (Whisper)
3. AI polishes — removes filler words, fixes repetitions, formats lists
4. Text appears in your target application

All processing happens on your machine. No cloud. No subscription. No data leaving your computer.

• 🎙️ Local transcription — Whisper.cpp, runs entirely offline
• ✨ AI text polishing — removes "um", "uh", "like", repetitions
• 🔒 100% private — audio never leaves your machine
• 🌍 100+ languages — speak in any language, auto-detected
• 🖥️ Works everywhere — any app, any text field
• 🤖 Local LLM — Ollama integration, or use OpenAI API
• 🌙 Dark mode — native look and feel

Early development. See REQUIREMENT.md for what's being built.

Menu Bar App (Tauri + React)
    │
    ├── Audio Capture (system microphone)
    │       │
    │       ▼
    ├── Whisper.cpp (local STT)
    │       │
    │       ▼
    ├── LLM Polish (Ollama or OpenAI)
    │       │
    │       ▼
    └── Text Injection (keyboard simulation)

Full system design in ARCHITECT.md.

Voix is built with an AI harness — the AI can write its own code, run its own tests, and improve itself within defined boundaries.

See AGENTS.md for the harness specification.

• macOS 12+ or Windows 10+
• Rust 1.70+
• Node.js 18+

# Clone the repo
git clone https://github.com/winsonwq/voxis.git
cd voxis

# Install dependencies
npm install

# Download Whisper model (first run will also prompt this)
# Recommended: "small" model for balance of speed/accuracy
./scripts/download-model.sh small

# Run in development
npm run tauri dev

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull a model
ollama pull qwen2.5:3b

# Start Ollama (runs on port 11434 by default)
ollama serve

Voix is an open project. Read AGENTS.md to understand the AI harness workflow, then:

1. Fork and clone
2. Create a branch: git checkout -b feature/your-feature
3. Make changes, run tests
4. Open a PR

For large changes, open an issue first to discuss direction.

MIT — do whatever you want with it.