DeepSeek AI experts, I am attempting to take some PDFs I have and produce a clean file (PDF, txt, epub, whatever) with the contained text. PDF Maths Translate as recommended in the OP looked promising but is laying the text over the original which doesn't achieve what I need (raw text will be far smaller in storage size than the originals which are unreasonably large scans). I assume this issue has already been solved by someone so does anyone have a good solution?