From file to sentences
First the book is split into chapters and paragraphs (see the format guide), then into short pieces a narrator voices in one go: a few sentences at a time. Every break falls between sentences, so a breath never lands mid-thought. Chapter titles get their own short announcement.
From sentences to speech
Today’s AI voices speak from context instead of stitching recorded syllables together. They hear the question mark, the aside and the list, and pace, pitch and emphasis follow. That is why a sarcastic line sounds sarcastic.
Book2Speech has two kinds of narrator. Standard voices read English, free and unlimited on every plan. Premium voices are richer and more expressive, and read all 42 languages. Together that is 265 narrators, with a range of ages, accents and styles.
What makes long-form work
•
Continuity: one narrator, one setting, from the first page to the last, with pieces joined seamlessly.
•
No waiting: narration is made just ahead of you, so a ten-hour book starts in seconds.
•
Structure: front matter, part dividers and back matter are recognised, so play lands on chapter one.
•
Listening tools: read-along highlighting, speed, sleep timer, bookmarks and your place synced between devices.
•
A file to keep: download the finished audiobook as M4B, MP3, Opus or WAV, with chapters and cover.
What to expect
•
Invented names and rare words are pronounced from the spelling, which usually sounds right and occasionally surprises you.
•
One narrator reads the whole book, dialogue included, the way many human narrators do.
•
Tables, equations and verse layout are read as text.
•
A scanned book is read exactly as its OCR text says, typos included.
Human narrators and AI
A great human narrator is still a performance worth paying for. AI narration wins on reach: the book you have, in the voice you choose, today, including the many books that will never get a studio recording. Drafts, papers, textbooks, small-press and out-of-print titles, and classics in translation are where it shines.