Full checklist
What an audiobook script is
In audiobook production, the script is simply the text the narrator reads, prepared for the ear. The audiobook script format for AI narration isn't a screenplay or a special markup: it's your manuscript with clear chapter headings, consistent dialogue punctuation and nothing in it that shouldn't be spoken. Get those right and chapters and speakers are detected correctly before you spend a single credit.
If you've wondered how to prepare a manuscript for audiobook narration, this page is the formatting reference: what FullCastBook's importer recognizes in each file type, how to set out dialogue, what to cut, and an audiobook script template you can copy. For the bigger picture of turning a finished book into audio, see manuscript to audiobook and the how it works overview.
Chapter headings the importer recognizes
Headings are the heart of the audiobook script format. In plain text, PDF and pasted text, a chapter starts at a short line on its own that reads like a heading: Chapter 7, CHAPTER SEVEN, Chap. 7, Part II, Book One or Section 4, or a word such as Prologue, Epilogue, Introduction, Preface, Foreword, Afterword or Interlude. A title can follow on the same line (Chapter 7: The Storm) or on the next short line. Chinese headings such as 第七章 or 第7回, and 序章, 楔子, 尾声 or 番外, work the same way.
A heading that is only a title (The Storm) or only a number (7) isn't recognized in plain text, so add the word Chapter. In Word, chapters start at your top heading level: use Heading 1 for chapter titles only, and the Title style rather than Heading 1 for the book's name (Word to audiobook has more). In EPUB, chapters follow the contents; in Markdown, the top heading level used twice. A contents list typed as headings with nothing under them is dropped.
Dialogue: quotation marks, tags and who's speaking
Dialogue is the other half of the audiobook script format. Speaker detection reads your dialogue the way a reader does, so help it the way you'd help a reader. Put spoken words in quotation marks and keep one style throughout: curly or straight double quotes in English, 「」 or “” in Chinese. Single-quote dialogue in the British style is harder to parse, because apostrophes look the same, so check those chapters' speakers more carefully.
Tag the speaker in long exchanges: a plain “said Mara” every few lines keeps attribution right, and so does an action beat (Mara set down her cup. “Fine.”). Start a new paragraph for each new speaker and never put two speakers in one paragraph. After import, check the speakers and fix any line given to the wrong character before you generate.
Scene breaks, italics and emphasis
Formatting doesn't carry into audio. Italics, bold and font changes are dropped on import, so an italic thought is read as narration and emphasis has to live in the words. If a thought should be in the character's voice, put it in quotes or reassign the line after casting; if one word must be stressed, reshape the sentence or add a short direction on that line, like [stressing “never”].
Scene breaks need care too. In Markdown, a --- line is dropped rather than read. In other formats, a row of asterisks (* * *) can come through as text; if you see one as a line in a chapter, delete it or replace it with an empty line. Never type “Scene break”, because the narrator will read it.
Front and back matter to skip
Audio doesn't need most of what surrounds the story. Cut or untick the copyright page, contents, also-by lists, acknowledgements you don't want read, image captions, indexes and, usually, footnotes. FullCastBook flags sections titled Copyright, Contents, Acknowledgements, Also by and similar as suggested skips. A long block before the first heading becomes an Opening section you can keep or drop; a few lines there are left out entirely.
Do add what the audiobook script format needs and print doesn't: short opening credits (title, author and a line saying the book is narrated with AI voices) and closing credits such as “The End”. Give each its own heading so it isn't lost. Each chapter's text starts with its heading, so the narrator announces “Chapter Seven: The Storm”; write headings the way you want them said.
Numbers, abbreviations and names
Write for the ear wherever a voice might guess: Dr. as Doctor, St. as Saint or Street, Louis XIV as Louis the Fourteenth, and 3:15 as a quarter past three if that's how the character would say it. URLs and email addresses read badly, so replace them with a sentence. Tables, charts and lists are read line by line; turn them into prose.
For invented names and foreign words, write the pronunciation into the text the voices read and keep a list. The audiobook pronunciation guide has the respelling method, and the fix for a single line later is just an edit and a regeneration.
Copy the audiobook script template
Save the lines below as a .md file, or paste them into Word, drop the # marks, give the first line the Title style and each ## line Heading 1. The first line becomes the book title and isn't read; every ## line starts a chapter and is announced.
# My Book Title
## Opening Credits
My Book Title, written by Author Name. This audiobook is narrated with AI voices.
## Chapter 1: The Letter
The letter came on a Tuesday. Mara read it twice before she spoke. “It's from Theo,” she said. “He's coming home.” Her brother didn't look up. “Then someone should warn the village,” said Ned.
## Chapter 2: The Harbor
## Closing Credits
The End. My Book Title was written by Author Name.
That is the whole audiobook script example: one heading per chapter, quoted dialogue with occasional tags, credits under their own headings. Upload it to the AI audiobook generator and check the chapter list before you cast.
