PDFMathTranslate

Why we included this project

Reading papers in a second language is a lot less painful with this tool. It takes a scientific PDF and returns a translated version that keeps the original layout, so formulas, tables, figures, and the table of contents stay where they belong instead of being flattened into plain text. That matters for math-heavy documents, where a naive text dump would scramble equations and make the translation hard to check against the source. You can plug in whatever translation backend fits your situation, from Google and DeepL to local models through Ollama or OpenAI-compatible services, which gives you a real choice between cost, privacy, and output quality. It also offers several ways to run it, including a command line, a GUI, Docker, and a Zotero plugin, so it can drop into an existing reading workflow without much fuss.