EPUB Translator: Create Bilingual Parallel E-books with LLM for a New Language Learning Experience

The open-source EPUB Translator project by Oomol Lab uses large language models to translate e-books, preserving the original text and generating a side-by-side bilingual version. It provides a new solution for language learners and cross-language readers, addressing the problem that traditional translation tools cannot meet the needs of in-depth reading.

EPUB Translator was born from an observation: pure machine translation easily loses the charm of the original text, while parallel reading is an effective method for foreign language learning. The project's core positioning is to combine the translation capabilities of large language models with e-book formats to create a new bilingual reading experience. Unlike replacement-based translation tools, it uses a side-by-side parallel presentation, allowing readers to see both the original text and the translation at the same time.

Technically, the project parses the internal structure of EPUB (compressed packages like HTML and CSS) to identify text areas and calls large language models for translation. Key highlights include:

1. Format preservation mechanism: Fully retains original typesetting, chapter structure, image positions, and metadata;
2. Bilingual parallel layout: Uses CSS to design two-column or staggered layouts, supporting readers to flexibly choose reading modes;
3. Multi-LLM integration: Supports multiple large language model APIs, allowing users to select models (such as GPT-4 or lightweight models) based on their needs.

• Language learners: Parallel reading helps establish vocabulary and grammar correspondences, and understand language usage in real contexts;
• Academic research: Quickly grasp the main idea of literature and avoid translation deviations by comparing with the original text;
• Cross-cultural reading: Experience the beauty of the original language while improving foreign language skills;
• Professional fields: Ensure accurate understanding of terminology in fields like law and medicine.

The usage threshold is low: Users need to prepare an EPUB format e-book, configure the large language model API key, start the translation process through a clear command-line interface, and it supports batch processing and custom translation parameters.

• Terminology library integration: Connect to professional terminology libraries to ensure translation consistency of vocabulary in specific fields;
• Translation memory: Establish personal translation memory libraries to reuse repeated content;
• Multilingual support: Expand to more minor languages;
• Reader integration: Deeply integrate with e-book readers to enhance the experience.

As an open-source project, EPUB Translator not only provides a practical tool but also demonstrates the idea of combining LLMs with traditional publishing processes. The code provides references for developers and inspires innovation. The project embodies the concept of technology serving people and is an example of AI integrating into daily study and work. Those interested can visit the GitHub repository to get the source code and documentation.