Turn a book or long document into audio
Drop in an EPUB, a PDF or your own manuscript and narrate it with a natural neural voice - one MP3 per chapter, at a pace you choose, free and with no credit card.
Long-form is genuinely a project rather than a single click, so the page below is honest about what that takes.
- One file per chapter
- EPUB and PDF
- Nothing uploaded
Edit the line to hear anything you like.
What a full book actually takes
Being concrete about this is more useful than a promise. At 7,000 characters per allowance and a two-minute refill:
| Length | Roughly |
|---|---|
| A short chapter (3,000 words) | About three renders, 20 minutes of audio, half an hour of your time |
| A novella (30,000 words) | Around 26 renders — an hour or two of steady work |
| A full novel (90,000 words) | Around 78 renders — an afternoon. Real, but very doable |
The wait is always the same flat two minutes, however hard you push - so a book is only ever a chapter, a short pause, and the next chapter.
How to structure it so the files are usable
- Split by chapter using blocks. Each block becomes its own MP3 under
parts/in the ZIP, numbered in order and tagged with its own title - so your player shows chapter names rather than one undifferentiated file. - Name the first line deliberately. The ID3 title is taken from it, so starting a block with “Chapter Four” gives you a sensible track name for free.
- Add a pause after each chapter heading. A second of silence does audibly what a page break does visually.
- Fix names once, early. Character and place names are exactly what a voice mispronounces. Add pronunciation respellings before you start rather than re-rendering chapters later.
EPUB is the better starting point
An EPUB stores real chapter structure and clean text, so it imports in order and without page furniture. A PDF stores positioned glyphs, which means running heads, footnotes and page numbers can land in the middle of a sentence and need clearing out first. If you have both, use the EPUB. Open the EPUB tool.
Questions
- Can it do a whole book in one go?
- No, and any site claiming otherwise is not being straight with you. One render covers up to 7,000 characters - about seven minutes, or a short chapter. A book is built chapter by chapter across several sittings, with the allowance refilling two minutes after you exhaust it.
- Do I get separate files per chapter?
- Yes, if you split the script into blocks. The ZIP download then contains one MP3 per block under parts/, each tagged with its own title - the dry narration, before any effects or music, which is what you want when assembling chapters.
- Is M4B supported?
- Not yet. M4B is the format with built-in chapter markers that audiobook players understand, and it is a fair thing to want. Today the output is MP3 files you can import into a player or combine yourself.
- What about copyrighted books?
- Converting a book you own for your own listening is generally personal use; distributing the result is not, and the voices are licensed for personal use only in any case. Public-domain texts are the safe ground for anything you want to share.
- Which voice suits long-form narration?
- Test before committing - a voice that sounds pleasant for thirty seconds can wear badly over an hour. The multilingual voices tend to hold up well across long passages. Generate a full chapter with your shortlist before doing the whole book.
- Does it handle the front matter and page numbers?
- EPUB files carry real chapter structure, so those import cleanly. PDFs are messier - headers, footers and page numbers can land mid-sentence. Everything arrives in the editor as plain text first, so you can strip the furniture before generating.