speech2text

Desktop speech-to-text powered by Groq's Whisper v3 API.
Record your voice, get the transcription pasted directly into whatever you're typing.

Install

winget install ozas.speech2text

Or download the latest installer from the releases page.

Set your Groq API key in settings
Hold your keybind (default Ctrl+Shift) or click the mic button
Speak
Release the keys (or click again) — transcription gets pasted into the active text field

A Dynamic Island-style overlay appears at the top of your screen showing recording/transcribing/done status.

Push-to-talk with configurable keybind (supports any key combo including modifier-only)
Real-time audio visualizer
Transcript history
Language selection (24 languages or auto-detect)
System tray with minimize-to-tray
Lightweight native app (~5MB)

Backend: Rust via Tauri v2 — Groq API, clipboard, raw Win32 keyboard hook, keystroke simulation
Frontend: React + Vite with Web Audio API visualizer
API: Groq Whisper Large v3

Requires Rust and Node.js.

npm install
npm run tauri dev

Release build:

npx tauri build

Produces a standalone NSIS installer in src-tauri/target/release/bundle/nsis/.