About epub2audiobook.com
A free EPUB to audiobook converter that runs in your browser. Here is who makes it, what it is made of, and how it stays free.
What it is
epub2audiobook.com turns an EPUB you own into an audiobook: a chaptered M4B with the book’s cover, or one MP3 per chapter. The narration is made by a neural speech model running inside the browser tab, so the book is read from your disk and never uploaded. There is no server-side code at all. The site is a set of static files.
It exists because the other options all ask for something. Server-based converters want the book, usually an account, and often a subscription. Command-line tools want Python, ffmpeg and an afternoon. This one wants a browser and about an hour of your computer’s time for a ten-hour book.
Not the GitHub project with the same name
There is an unrelated open-source command-line tool called epub2audiobook on
GitHub, which sends the text to Microsoft Edge’s online speech service. This site
isn’t that, isn’t a hosted version of it and shares no code with it. If you
read a description of “epub2audiobook” that mentions Edge TTS, Python or
Russian-language books, it is about that project.
The tool comparison covers it alongside the other converters, ours included.
Who makes it
An independent developer, as a side project. No company, no investors, and no data business behind it, because there is no data. The same approach (do the work in the browser and keep the files on your machine) is the one the site applies to everything it does.
Who pays for it
Running it costs almost nothing, which is the only reason it can be free without a catch. The pages are served by Cloudflare Pages on its free tier. The voice models live in Cloudflare R2, which charges nothing for downloads. The domain is a few dollars a year.
The card at the side of the page is our sponsor, Talisim, a travel eSIM company, and it helps keep the site free. It is plain markup and a link, and it loads nothing from their servers; the privacy page has the details.
Built on
The hard parts are other people’s open-source work, and the site would not exist without it.
- Piper: The speech model, and the 11 voices.
- ONNX Runtime Web: Runs the model as WebAssembly, across your processor’s cores.
- eSpeak NG: Turns text into phonemes for Piper, via the phonemizer package.
- Mediabunny: Encodes the AAC and MP3 audio and writes the M4B container.
- fflate: Reads the EPUB, which is a zip, and writes the MP3 zip.
- Astro: Builds the site as static pages.