About paper2epub

paper2epub turns scientific papers in PDF into EPUBs that read well on a Kindle, a Kobo or a phone. It’s free, it needs no account, and it’s made and run by one person.

Who makes it

I’m Miguel García, a data scientist and software developer from Spain. I read a lot, mostly science fiction and fantasy, and nearly all of it on an e-reader. Papers were the exception: they stayed on my laptop, because a two-column PDF on a six-inch screen is all zooming and losing your place.

So I wrote the tool I wanted. I write the converter, run the server and write the blog myself. On my personal blog, migue8gl.dev, I write about machine learning and whatever else I find interesting.

Why a converter just for papers

A PDF fixes every line in place for a printed page. An e-reader can only shrink that page or let you pan around it. General-purpose converters, Calibre included, read a PDF as a stream of lines, which works for a novel and falls apart on a paper: equations come apart symbol by symbol, tables turn into loose words and the two columns get mixed up. arXiv’s HTML versions help, but only for some papers. paper2epub is built for what papers have: columns, equations, tables, figures, footnotes and citations. The blog has the comparison with Calibre and every other way of reading a paper on an e-reader, tested on the same pages.

What you get

The paper rebuilt as a real e-book. The text flows to fit your screen and your font size, with its italics, subscripts and superscripts. Equations are typeset with LaTeX and stay crisp at any size, tables stay tables, figures keep their captions, and every citation, footnote and cross-reference is a link you can tap. Each section is a chapter, with a table of contents.

The text of your EPUB is the text of the paper: nothing is summarised, translated or rewritten.

How it’s tested

Every change to the converter is checked on a set of real papers before it reaches this site, and the EPUBs pass epubcheck, the reference validator for e-books, with no errors.

It doesn’t get everything right. Papers with unusual layouts can still come out with a piece in the wrong place, and a scanned PDF, which is only pictures of pages, can’t be converted at all. If a paper comes out badly, tell me which one, and it helps the next version.

The blog

The blog is about reading scientific papers on e-readers in general: guides for Kindle and Kobo, which device to buy, and comparisons with the other tools. The comparisons are real tests: the same pages, converted by each tool, shown on the same screen, with the versions and settings used.

How it’s paid for

The site is free and there is no paid plan. The server is paid for by the ads on its pages and by readers who buy me a coffee. Ads never go inside the books. What happens to your files and data is in the privacy page, and the rules of use are in the terms.

Contact

Write to me at migue8gl@gmail.com: a paper that converts badly (an arXiv link is best), an idea, a mistake in a post or a question about your data. I read everything and answer as soon as I can.