Turn a PDF or EPUB into an audiobook
Chapters are found for you, page numbers and running headers are cleaned out, and a natural voice reads each chapter on your own device. Download chapter by chapter. Free, no sign-in, nothing uploaded.
Drop a book here
PDF, EPUB, DOCX, TXT, Markdown or HTML
Book
How it works
- Read. The file is opened in this tab. PDFs give their text layer and bookmarks, EPUBs their chapter files and table of contents.
- Split and clean. Chapters come from the best evidence in the file. Page numbers, running headers, footnote marks and links are taken out; you can see and edit what will be read.
- Voice. Kokoro, an open speech model, runs on your graphics chip. It downloads once (about 110 MB) and stays in this browser.
- Keep. Every finished paragraph is saved on this device, so you can pause, close the tab and continue later. Download chapters as you go.
Questions
Is my book uploaded?
No. The file is read inside this tab and the voice runs on your device. The only downloads are the voice model and voices, once, from SkillSafe's model host. Your text is never sent anywhere.
How long does a book take?
On a laptop with a recent graphics chip the voice renders about ten times faster than it speaks, so a ten-hour book takes about an hour. A live estimate appears after the first paragraph. Keep the tab open; it can sit in the background.
My browser says the voice cannot run here.
The natural voice needs WebGPU: recent Chrome or Edge on a computer, or Safari 26. Without it you can still clean the text, split chapters and listen with your browser's built-in voice, but that voice cannot be saved as a file.
Why Opus and not MP3?
Your browser can encode Opus itself, quickly and without extra code, and an Opus file is smaller than an MP3 of the same quality. Every current phone, browser and VLC plays it. Pick WAV if you need a file to edit.
Can it read scanned PDFs or other languages?
It needs a PDF with a text layer; a scan is detected and named. The voices are English (American and British).