How to format a manuscript for narration

Prepare means fixing the manuscript before anyone reads it aloud. Messy files slow narration: wrong breaks, unclear speakers, names the engine guesses at. You end up re-recording or patching later.

This chapter covers format, dialogue, chapter marks, and what to skip. You'll walk out with one clean manuscript, a short pronunciation list if you need it, and files labeled so production can start without back-and-forth.

Send plain text, Word, or a Google Doc. One column. Strip anything meant for print, not listening.

• Remove running headers, footers, page numbers, and columns
• Drop drop caps, text boxes, and decorative fonts
• Use consistent chapter headings (same wording and level every time)
• Keep body text in one font; bold and italic are fine

If you exported from InDesign or a template, paste into a fresh doc and simplify before you upload.

Dialogue has to read clearly on its own. Mark who's speaking when it's not obvious from context.

• Start a new paragraph for each speaker turn
• Use quote marks consistently; avoid burying dialogue mid-paragraph
• For invented names and tough proper nouns, add a phonetic line in your notes: Kehlani (keh-LAH-nee)
• Note accents only when the read should change (a character code-switches, a place name isn't English)
• Spell out abbreviations that sound odd when read literally (Dr., St., etc.) or flag them in your style guide

You don't need IPA. A simple phonetic hint in a separate list is enough.

One manuscript file is easiest. If you split by chapter, use the same break convention in every file.

Mark scene breaks with something plain: a line with ***, or a blank line your producer already knows about. Skip ornaments (flourishes, dingbats, unicode dividers). TTS treats them as random characters or long pauses.

Chapter titles should match your file names later. "Chapter 12" and "Twelve" are both fine; pick one pattern and stick to it.

Keep a short living doc alongside the manuscript. Update it when you add names or change how something should sound.

Include:

• Character and place names with phonetic spellings
• How you want numbers, dates, and times read ("twenty twenty-four" vs "two thousand twenty-four")
• Words you always want the same way (made-up terms, titles, recurring jargon)

A page or two is plenty. Send it as a separate file when you upload.

Not everything in the book belongs in the audiobook. Decide now what gets read and what gets cut.

• Usually skip: half-title, copyright page, ISBN blocks, ads for other titles
• Optional: dedication, epigraph (read only if you want it in the audio)
• Often at the end: about-the-author, acknowledgments

Opening credits ("This is… narrated by…") are built separately at publish time. Don't paste them into chapter one unless you mean them to be read as part of the story.

Name files so a human can sort them: 01-prologue.docx, 02-chapter-one.docx, or one manuscript-full.docx. Match whatever chapter labels are inside the text.

Keep one source of truth. Send the final manuscript, not a draft with track changes or comments still open. If you use Word or Google Docs, resolve suggestions before export.

Upload the manuscript plus your style guide if you have one. If anything moved since the last version, say so in the cover note.

Next is narrate: pick a voice, set pacing, and get through the read.