Translate Books With LLMs

Translate Books With LLMs translates entire books, documents and subtitles while retaining their formatting. It's for readers and translators who need to work through long texts with a local LLM or a cloud model. The app runs on Windows, macOS and Linux through a browser interface, and supports self-hosted Docker deployment.

EPUB output retains the book's styles and structure, while SRT translations keep the original subtitle timing. DOCX and TXT are supported too. The app handles long documents in segments while keeping context between them, and saves progress so interrupted translations can continue without starting over.

Glossaries help keep names and terminology consistent across a book. Writing style presets can come from sample books or your own instructions, giving the translation a consistent register and rhythm. The app can also derive a glossary and style from the document itself. Optional OCR and typography cleanup addresses source text issues, and a second translation pass focuses on literary polish.

With Ollama, document content stays on your machine. Local OpenAI-compatible backends include llama.cpp, LM Studio, vLLM and LocalAI. Cloud translation sends text to the selected provider; supported services include OpenAI, Gemini, Mistral, DeepSeek, NVIDIA NIM, OpenRouter and Poe, which require API keys.

The project is open source under AGPL-3.0. A command-line interface supports automated translation work. A self-hosted server shares translation progress, completed files and job history across connected browsers in one workspace, without separate user accounts.