Speaky

A cross-platform voice input application that converts speech to text in real-time. Press a hotkey, speak, and your words are automatically typed at the cursor position.

English | 中文

Features

• Global Hotkey - Press Alt+Space (or custom shortcut) to start/stop recording from anywhere
• Real-time Transcription - See your speech converted to text instantly
• Auto-paste - Transcribed text is automatically typed at cursor position
• LLM Post-processing - Optional AI polish to add punctuation, fix errors, and improve formatting
• Multi-provider Support - Works with DeepSeek, OpenAI, Kimi, Gemini, Zhipu, Ollama, and any OpenAI-compatible API
• Recording Indicator - Visual indicator shows when recording is active
• History - Browse and copy previous transcriptions
• Dark/Light Theme - Beautiful UI with theme support
• Cross-platform - Works on Windows, macOS, and Linux

Screenshots

Installation

Pre-built Binaries

Download the latest release for your platform from the Releases page.

• Windows: Download .msi or .exe installer
• macOS: Download .dmg file
• Linux: Download .AppImage or .deb package

Build from Source

Prerequisites:

• Node.js 18+
• Rust 1.70+
• Tauri CLI

# Clone the repository
git clone https://github.com/guangzhaoli/Speaky.git
cd Speaky

# Install dependencies
npm install

# Run in development mode
npm run tauri dev

# Build for production
npm run tauri build

Configuration

ASR (Speech Recognition)

Speaky uses Volcengine Doubao ASR for speech recognition. You'll need to:

1. Create an account at Volcengine Console
2. Enable the Speech Recognition service
3. Create an application and get your credentials:
   • App ID
   • Access Token
   • Secret Key (optional)

Enter these in Settings > General > API Configuration.

LLM Post-processing (Optional)

Enable AI-powered text polish in Settings > LLM Polish:

1. Toggle "Enable LLM Post-Processing"
2. Choose a processing mode:
   • General - For everyday text input
   • Code - Preserves technical terms and syntax
   • Meeting - Formal writing style
3. Add and configure an API provider

Supported providers:

• DeepSeek
• OpenAI
• Kimi (Moonshot)
• Google Gemini
• Zhipu (GLM)
• Ollama (Local)
• Any OpenAI-compatible API

Usage

1. Start Recording: Press the global hotkey (default: Alt+Space) or click the microphone button
2. Speak: Talk naturally - your speech will be transcribed in real-time
3. Stop Recording: Release the hotkey or click the button again
4. Auto-paste: The transcribed text is automatically typed at your cursor position

Keyboard Shortcuts

Shortcut	Action
Alt+Space	Start/Stop recording (customizable)

Tech Stack

• Framework: Tauri v2 (Rust + React)
• Frontend: React 19 + TypeScript + TailwindCSS v4
• Audio: cpal (cross-platform audio capture)
• Keyboard: enigo (keyboard simulation)
• ASR: Volcengine Doubao ASR 2.0 (WebSocket binary protocol)
• LLM: OpenAI-compatible API

Project Structure

speaky/
├── src/                    # React frontend
│   ├── App.tsx            # Main application component
│   ├── main.tsx           # React entry point
│   └── style.css          # TailwindCSS styles
├── src-tauri/             # Rust backend
│   ├── src/
│   │   ├── lib.rs         # Tauri application setup
│   │   ├── commands.rs    # IPC commands
│   │   ├── state.rs       # Application state
│   │   ├── audio/         # Audio capture
│   │   ├── asr/           # Speech recognition
│   │   ├── input/         # Keyboard simulation
│   │   └── postprocess/   # LLM post-processing
│   └── Cargo.toml         # Rust dependencies
├── package.json           # Node.js dependencies
└── tauri.conf.json        # Tauri configuration

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Tauri for the amazing framework
• Volcengine for the ASR service