Voice Type

Voice Type is a desktop push-to-talk speech-to-text app built with Rust, Tauri, React, and Bun.

Hold your hotkey, speak, release, and the app transcribes your speech with Groq Whisper and types the result into the active window.

Current State

• Version: 0.7.2
• Desktop shell: Tauri 2
• Frontend: React + TypeScript + Vite
• Backend engine: Rust + Tokio
• Package manager: Bun

Features

• Push-to-talk transcription
• Global hotkey capture
• Floating recording HUD
• Appearance customization
• Optional grammar cleanup
• Text cleanup filters
• Local config storage

Project Layout

.
|- Cargo.toml
|- src/                  # Rust backend engine
|- frontend/             # React desktop UI
|- src-tauri/            # Tauri shell and command bridge
|- package.json
`- assets/

Development

Prerequisites

• Rust 1.85+
• Bun
• Windows: WebView2
• Linux: Tauri/WebKitGTK prerequisites
• macOS: Xcode command line tools

Commands

npm install
npm run tauri:dev

Useful checks:

npm run build
cargo check --lib
cargo check --manifest-path src-tauri/Cargo.toml

Configuration

Config is stored at:

• Windows: %APPDATA%\voice-type\config.json
• macOS: ~/Library/Application Support/voice-type/config.json
• Linux: ~/.config/voice-type/config.json

You can either:

• put the API key in the app settings, or
• leave api_key empty and use GROQ_API_KEY

Notes

• This repo is the current codebase.
• The older Python-based implementation should be treated as legacy.
• Bundled public image sources are documented in ASSET_SOURCES.md.

Building for Release

NSIS Installer

npm run tauri:build

Output: src-tauri/target/release/bundle/nsis/Voice Type_0.7.2_x64-setup.exe

Microsoft Store (MSIX)

Requires Windows SDK with makeappx.exe (already installed).

npm run tauri:build
npm run msix

Output: msix-output/Boring877.VoiceType_0.7.2.0_x64__kg07y93afj4jj.msix

Upload the .msix file to Microsoft Partner Center to update the Store listing.

A symlink to the MSIX output folder is on the Desktop (voice-type-msix) for quick access.