Drop an audiobook without chapters: it finds where the narrator announces each one, titles it from what was said, and writes the chapters in without re-encoding. Nothing is uploaded.
Finds where the narrator says “Chapter …” and sets chapters there · Whisper runs in your browser · nothing is uploaded
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 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.
The first transcript downloads a Whisper speech model once (about 43 MB for the quick English one), faster with a graphics card (WebGPU).
How to use it
One long MP3, M4B or M4A whose chapters are missing or wrong.
The book is measured to find its long pauses, then Whisper listens to the first seconds after each one.
Every spoken heading becomes a chapter with its title. Untick, rename or play any of them.
Download the audiobook with the chapters written in (the audio isn’t re-encoded), or a chapter list.
Transcribing a whole ten-hour book takes a while even on a fast computer. Chapter announcements almost always follow a pause of two or three seconds, so the detector listens only there: a typical book has a few hundred long pauses, and seven seconds of each is about half an hour of audio instead of ten hours. Need the whole text? Use the audiobook transcriber.
FAQ
Narrators pause before each chapter and announce it: “Chapter Twelve. The Storm.” The detector measures the whole book to find its long pauses, then transcribes only the seven seconds after each one with Whisper. When it hears a heading (Chapter, Part, Book, Prologue, Epilogue, or a bare number), it sets a chapter there and uses what was said as the title.
Then there is nothing to hear. Use split by silence to put chapters at the longest pauses, or the chapter editor to place them by hand.
No. Saving rewrites only the chapter list, copying the audio as it is, so there is no quality loss and it takes seconds.
No. Everything runs in your browser; the speech model downloads once and is kept in your browser’s cache.
Read the guide
Keep going