subtitle-translator: Batch-Translate Subtitle Files

Overview

Subtitle Translator batch-translates .srt, .ass, .vtt, and .lrc files through traditional machine-translation services or LLM providers. It separates dialogue from timing and file structure before sending text for translation, then restores the original structure in the output.

It is aimed at people translating many subtitle files, such as a full season, as well as users who want bilingual output or a scriptable workflow. The web app runs in the browser, and a CLI uses the same translation engine for headless runs.

Key Features

• Batch-process multiple subtitle files and translate to several target languages in one pass.
• Supports SRT, ASS, VTT, and LRC, with options for bilingual output and format conversion.
• Connects to traditional translation APIs and LLM providers, including custom OpenAI-compatible endpoints.
• Preserves timecodes and cue structure by extracting dialogue locally before translation.
• Offers context-aware translation for LLMs and local caching for translated text.
• Provides a CLI for scripted translation, with resumable runs and settings exported from the web app.
• Keeps subtitle parsing and API-key storage in the browser; requests go to the selected translation endpoint.

Use Cases

• Subtitle editors can translate a season of episodes without manually processing each file.
• Fansub and localization teams can produce bilingual subtitles or exports in multiple languages.
• Video creators can translate SRT or VTT captions while retaining the original timing.
• Developers can script repeatable translation jobs with the CLI or connect a local model endpoint.

Getting Started

Try the web app, or run the project locally:

yarn install
yarn dev

See the repository README for CLI usage, provider setup, and deployment details.

Considerations

• Translation quality and cost depend on the selected service; LLM usage may incur token charges, while free services can be rate-limited.
• Some providers require API keys or relay configuration. Requests are made to the configured provider, so browser-based use depends on that endpoint being accessible.
• Local development and the CLI require Node.js 24 or later, according to the README.