Real-time Speech Translation

more than 50% of the code is AI generated

A real-time speech-to-text and translation application with Picture-in-Picture support.

Features

• 🎤 Real-time speech recognition (Japanese)
• 🌐 Live translation (Japanese to en/zh/ko)
• 🖼️ Picture-in-Picture mode for overlay display (modern chromium only)
• 🔊 Multiple microphone support

Setup

1. Install dependencies:

pnpm install

2. Get your Soniox API key:

   • Visit Soniox to create an account and get your API key
   • The API key will be securely stored in your browser's localStorage

3. Start the development server:

pnpm dev

Tech Stack

• React 19 + TypeScript
• Vite
• Tailwind CSS
• Soniox Speech-to-Text API
• Document Picture-in-Picture API

Privacy

⚠️ Important: Your API key is stored locally in your browser and never sent to any server except Soniox's API. Make sure to keep your API key confidential.

License

MIT