Real-time speech translation is addressed by this desktop application, targeting users on macOS and Windows who require live transcription and translation without relying on intermediary servers. It offers a privacy-conscious approach by utilizing user-provided API keys and integrates multiple free and paid Text-to-Speech (TTS) options, providing a flexible and low-latency solution.

How It Works

The application is built with Tauri, leveraging Rust for the backend and a WebView for the frontend. It captures audio directly from system sources (macOS ScreenCaptureKit, Windows WASAPI) or microphones (cpal). This audio is sent to the Soniox API for Speech-to-Text (STT) and translation, achieving approximately 2-3 seconds of latency. Translations are then displayed in a minimal overlay UI, with optional TTS narration provided by integrated services.

Quick Start & Requirements

• Install: Clone the repository, install Node.js dependencies, and build the Tauri application:

  git clone https://github.com/phuc-nt/my-translator.git
  cd my-translator
  npm install
  npm run tauri build
  

• Prerequisites: Rust (stable), Node.js 18+, macOS 13+ or Windows 10+.
• Links:
  • macOS Installation (EN): https://github.com/phuc-nt/my-translator/blob/main/docs/install/macos-en.md
  • Windows Installation (EN): https://github.com/phuc-nt/my-translator/blob/main/docs/install/windows-en.md
  • TTS Guide (EN): https://github.com/phuc-nt/my-translator/blob/main/docs/tts/en.md

Highlighted Details

• Supports over 70 source languages with real-time translation and ~2-3s latency.
• Offers three TTS providers: free Edge TTS, free-tier Google Chirp 3 HD, and premium ElevenLabs, with speed control available for Edge and Google.
• Features a dual-panel view for source/translation side-by-side, smart auto-scrolling, and quick font size adjustments up to 140px.
• Enables custom translation terms for domain-specific vocabulary (e.g., medical, religious).
• Includes an experimental local mode for Apple Silicon Macs using MLX, Whisper, and Gemma for on-device translation (JA/EN/ZH/KO → VI/EN).
• Prioritizes privacy with no intermediary servers, local API key storage, and zero telemetry.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord/Slack), or roadmap are present in the provided README.

Licensing & Compatibility

The project is released under the MIT License, permitting commercial use and integration without significant restrictions.

Limitations & Caveats

The core functionality relies on external APIs (Soniox, Google Cloud TTS, ElevenLabs), incurring costs beyond the free tiers. The experimental local mode is restricted to Apple Silicon hardware.