BabelDOC: Yet Another Document Translator
English
BabelDOC is an open-source, AI-powered document translator developed by the funstory-ai team, specifically designed for translating PDF documents while preserving their original formatting and layout. It is particularly optimized for academic papers and professional documents.
Key Features
- Accurate Typesetting Restoration: The tool utilizes advanced layout analysis technology to retain the original document's intricate details, such as fonts, colors, spacing, and overall structure.
- Intelligent Formula Processing: Ensures that mathematical formulas are perfectly preserved and integrated within the translated text.
- Bilingual Comparison: Users can generate bilingual PDFs that display the original and translated text side-by-side or on alternating pages, facilitating easy comparison and improved reading efficiency.
- Multi-Engine and LLM Support: Integrates with popular translation services like Bing Translate and OpenAI, and is compatible with large language models such as GPT-3.5 and GPT-4. Also supports flexible integration with OpenAI-like interfaces and local models like Ollama.
- Batch Processing: Allows simultaneous translation of multiple PDF files, catering to large-scale document translation needs.
- Open Source: Hosted on GitHub, encouraging community contributions and development.
Technical Approach
BabelDOC is available as a beta version with a free usage quota through Immersive Translate - BabelDOC. For local deployment, it can be used with PDFMathTranslate-next, which offers command-line and graphical user interfaces. A Python API is also available for developers to embed its functionalities into other applications.
Known Limitations
- Parsing errors in author and reference sections
- Limited support for lines
- Potential skipping of large pages
About funstory-ai
funstory-ai was established in November 2017. The company specializes in creating AI-oriented solutions for promoting content, particularly literature and information, internationally through its AI-assisted translation and production system.
GitHub: https://github.com/funstory-ai/BabelDOC
中文
BabelDOC 是由 funstory-ai 团队开发的开源 AI 文档翻译工具，专为 PDF 文档翻译设计，同时保留原始格式和布局，针对学术论文和专业文档做了深度优化。
核心功能
- 精确排版还原：采用先进布局分析技术，保留原文档的字体、颜色、间距和整体结构等细节。
- 智能公式处理：确保数学公式在译文中完美保留并与文字无缝整合。
- 双语对照：用户可生成双语 PDF，左右对照或交替页面显示，便于比较阅读。
- 多引擎与大模型支持：集成 Bing Translate、OpenAI 等主流翻译服务，兼容 GPT-3.5/GPT-4 等大语言模型，并支持 OpenAI 兼容接口和 Ollama 等本地模型。
- 批量处理：支持同时翻译多个 PDF 文件，满足大规模文档翻译需求。
- 开源：项目托管于 GitHub，鼓励社区贡献与开发。
技术方案
BabelDOC 可通过 Immersive Translate - BabelDOC 免费使用 beta 版本。本地部署可使用 PDFMathTranslate-next，提供命令行和图形界面。开发者也可通过 Python API 将其功能嵌入其他应用程序。
已知局限
- 作者和参考文献区域可能存在解析错误
- 对线条的支持有限
- 可能跳过大型页面
关于 funstory-ai
funstory-ai 成立于 2017 年 11 月，专注于通过 AI 辅助翻译与生产系统，将文学和信息内容国际化推广的 AI 解决方案。