EPUB to Audiobook: free AI narration, M4B with chapters

Where it runs

Works in a current Chrome, Edge, Firefox or Safari, on Windows, Mac, Linux and Chromebook computers, Android phones, iPhones and iPads. Nothing to install, no account.

The first narration downloads the voice model once: about 93 MB on a processor, or 327 MB with a graphics card (WebGPU, in current Chrome and Edge), which is several times faster.

The first file loads the audio engine (ffmpeg, about 31 MB) once, and your browser keeps it. Very long books are easiest on a computer, which has more memory than a phone.

4 simple steps.

Drop your EPUB

A DRM-free EPUB you own or have the rights to. Its chapters become the chapter list; untick the copyright page and other front and back matter.

Pick a voice

Ten natural voices, American and British. Tap an orb to hear it, set the speed, and hear a paragraph of your own book.

Let it narrate

Chapters are voiced one after another in your browser, saved as they finish. Listen to finished chapters while the rest generate.

Download

An M4B with chapters and the EPUB’s cover, or one tagged MP3 per chapter.

How it works

The EPUB’s spine and headings become chapters. Each paragraph is tidied for speech (abbreviations, numbers, footnote markers and web addresses), turned into phonemes with open pronunciation dictionaries, and voiced by Kokoro-82M, an Apache-licensed text-to-speech model, in a background worker on your graphics card (WebGPU) or processor. Every few minutes of audio is levelled, encoded and saved in your browser, so a closed tab costs at most a few minutes. The finished chapters are joined without re-encoding into an M4B with chapter marks, title, author and the EPUB’s cover.

Where it runs	Download (once)	Speed
Graphics card (WebGPU)	about 330 MB	usually faster than real time
Processor (WebAssembly)	about 93 MB	about 2–3 × the audio’s length

Have the book as a PDF instead? PDF to audiobook does the same with its page numbers, headers and footnotes taken out. To audition the voices on a passage first, or tune a name’s pronunciation, use Kokoro TTS.

Tip: run long books through the manuscript analyzer first to see the chapter list and reading time, and the pronunciation guide maker to collect unusual names.

Questions, answered.

Is my book uploaded anywhere?

No. The EPUB is read in your browser and the AI voice (Kokoro, an open-source model) runs on your device. The voice model downloads once from Hugging Face and stays in your browser’s cache; your text never leaves your computer.

How long does a whole book take?

It depends on your device. With WebGPU (current Chrome and Edge on most computers) narration usually runs faster than real time. On a processor alone it takes roughly two to three times the length of the audio, so a 7-hour book can take most of a day: you can pause, close the tab and resume later, because finished parts are saved in your browser.

Can I narrate a Kindle or Apple Books purchase?

No. DRM-protected EPUBs are declined, and you need the rights to turn a text into audio. This tool is for your own manuscripts, public-domain books and texts you have permission to narrate.

Can I sell an audiobook made with an AI voice?

The voices are free to use, including commercially, under the Apache 2.0 licence. Retailers set their own rules: ACX (Audible) doesn’t accept AI narration unless Audible has authorised it; Kobo and Spotify for Authors accept AI-narrated audiobooks with a label; Google Play and Apple Books offer their own AI narration. Our AI audiobooks guide lists each store’s rules.

Which languages are supported?

English for now, in American and British voices. Pronunciation comes from open dictionaries of about 180,000 words plus rules for names and numbers; use the pronunciation guide to check unusual names.

Why does the copyright page show up unticked?

Front and back matter (copyright, contents, acknowledgements, “also by”) is detected by its title and left out by default. Tick anything you want read.