Hear your book.In more than one voice.

Why authors listen to their own books

Reading your manuscript aloud is one of the oldest editing tricks there is, because the ear is a harsher critic than the eye. You hear the word you used three times in a paragraph, the rhythm that goes flat, the line of dialogue that sounds written rather than spoken. Most authors know this and few do it for a whole book, because reading three hundred pages aloud takes days.

Narration lets you listen instead: on a walk, on the commute, with the manuscript open and a pen in hand. Play from any paragraph, slow it down for the passage you’re unsure of, and set a sleep timer for the chapter you want to hear before bed. It is a different way of meeting your own book, and it tends to find problems that a fifth silent read misses.

A narrator and a cast

Choose a narrator’s voice from the previews, then give your characters voices of their own. If you have a story bible, ✦ Suggest a cast reads each character’s entry and proposes a voice and a standing instruction, keeping the narrator and the main characters distinct. Nothing is assigned until you confirm it, and you can pick every voice yourself instead.

When a chapter is narrated, a director pass splits it into segments, each with a speaker and a delivery. Dialogue goes to the character’s voice and the tag that follows goes back to the narrator. In a first-person chapter, the narrator is the point-of-view character. The words themselves are never changed: the text that is spoken is checked against your manuscript, and if it can’t be matched it falls back to the narrator reading paragraph by paragraph.

Casting a book for the ear follows the same rule as casting a radio play: the voices listeners must tell apart in a conversation should sound clearly different. Keep the narrator and your main characters distinct, and let minor characters share voices, told apart by their standing instructions (older, clipped, nervous, amused). A cast that is too large blurs; a few well-chosen voices carry a whole ensemble.

Direction without markup

You don’t need to learn any tags or scripting to direct the narration. The project’s narration settings hold the narrator, the cast, how strongly it is acted (subtle, natural or dramatic), the speed, whether chapter headings are read, and the length of the pauses after paragraphs, headings and scene breaks. Scene breaks become a longer pause on their own.

Because these settings belong to the project, two collaborators can’t accidentally give one book two different narrators. Anyone who can read the project, including commenters and viewers, can listen; only those who can change the project’s rules and formatting can change the voices. That makes narration a handy way to share a draft with a co-author or a trusted reader who would rather listen than read.

• Voice previews before you choose
• Intensity: subtle, natural or dramatic
• Speed, headings read or skipped, and pause lengths

Paying only for what changed

Narration is an AI action, so it shows what it will cost before it runs, for one chapter or a whole book. Narrating a book skips the chapters whose narration is already current. The work runs in the background, with its progress in the bell and a Stop button, so you can keep writing while a book is being read.

Authors revise, and narration keeps up without starting over. Change two paragraphs and only those two are narrated again and billed. Unchanged passages are never paid for twice, even on another branch or in an alternate that shares the text. The chapter’s details page shows when paragraphs have changed since the last narration, so you know when it’s worth updating.

A good rhythm is to narrate once a chapter is in decent shape rather than after every sentence: listen, mark what jars, revise, and update only the paragraphs you changed. Each later pass pays only for what you changed, and you hear the chapter improve.

A real player, in the studio and on the page

In the studio the player sits at the foot of the page: play and pause, skip back and forward, speed, a sleep timer, the contents, lock-screen controls on a phone, and play from any paragraph. The text follows the narration until you scroll or type. Every chapter can be downloaded as one MP3, and narrations can be included when you export the project.

If you publish the story, readers can listen on the chapter page too. The audio is offered only when the narration matches the published text exactly, so the voice never reads words the page doesn’t show. The player always says “AI narration”, so listeners know what they are hearing. Readers don’t get a download or a saved position; the MP3 files stay yours.

What it is, and what it isn’t

This is text-to-speech narration for drafting, revising and sharing with your own readers. It isn’t a recording-studio audiobook, and it isn’t prepared to meet the technical or content requirements of any audiobook retailer or distributor. Many of them have their own rules about AI narration, so check before you treat these files as anything more than a listening copy.

If you plan a commercial audiobook later, narration still earns its place. It’s a quick way to hear how your dialogue sounds, to find the tongue-twisters and the run-on sentences a human narrator will stumble on, and to notice the scenes where it’s unclear who is speaking. Those are cheaper to fix in the manuscript than in a recording booth.