Overview

Nola Translator turns speech into real-time captions and translations. Run speech recognition and translation locally, use cloud services, or combine the two.

Features

• Local models: Run recognition models such as Qwen3-ASR and SenseVoiceSmall, and translation models such as HY-MT2 locally. The app can also search for and install other models from Hugging Face.
• Cloud models: Use speech recognition from Aliyun, MiniMax, and other providers; connect translation models through OpenAI-compatible, Anthropic, or Ollama APIs; or use the free MyMemory translation service.
• GPU acceleration: Detects available compute devices and lets you configure separate devices for recognition and translation.
• Browser extension: Includes a browser extension for Chrome and Edge that displays bilingual captions on video pages. You can also configure desktop system audio as an input source.
• Real-time caption overlay: Supports system output, a selected audio device, or a microphone. Captions appear in a separate overlay, where you can adjust its position, font size, colors, opacity, and original/translated text layout.
• Flexible configuration: Install models as needed and choose recognition and translation methods, language directions, compute devices, and caption styles. API credentials are securely stored by the app.
• History and export: Save caption history and export it as TXT, SRT, or WebVTT. Audio recording can be enabled optionally.

Interface Preview

Installation

• Download the latest release

Windows x64 is supported. You can choose the EXE or MSI installer. Support for other platforms is planned once development has stabilized.

• Online documentation

Language Support

[!TIP] The languages available for recognition and translation depend on the selected model's capabilities. The tables below show the capabilities of the recommended models included in the app. If you have a model to recommend, submit it as an issue; we will evaluate it for possible inclusion. For first-time users, we recommend SenseVoiceSmall for recognition and HY-MT2 1.8B for translation.

Recognition Models

Translation Models

Cloud Models

LLM API Configuration

[!TIP] The LLM protocols here are used for translation. Aliyun and MiniMax ASR models use separate connection settings and credentials under Audio and Recognition. Local models require no API configuration.

Supported Protocols

The app supports the following API protocols:

Configuration Examples

Configuration Steps

1. Select Cloud Translation in the translation settings.
2. Choose the API format supported by your provider.
3. Enter the API URL, API key, and model ID.
4. Adjust the context window and maximum output tokens as needed.

Project Structure

assets/brand/                  Logo and brand assets
docs/                          VitePress product website and multilingual documentation
engine/nola_translator_engine/  Python audio, recognition, and translation engine
engine/tests/                   Python tests
scripts/                        Setup and build scripts
src/main/                       Electron main process
src/preload/                    IPC bridge
src/renderer/                   React interface
src/shared/                     Shared contracts and settings
tests/                          Application tests

Development and Build

npm run typecheck
npm test
.\.venv\Scripts\python.exe -m pytest engine\tests
npm run build

Build the Windows x64 installer:

npm run dist:win

Contributing

Fork the repository, create a branch, make your changes, and open a pull request. Briefly describe your changes and the checks you ran. For design and code conventions, see DESIGN.md and AGENTS.md.

License

Nola Translator's original code is licensed under GNU GPL v3.0 only. Third-party components and models are subject to their respective licenses; see Third-Party Notices.