Audiobook Chapter Workflow: From Manuscript to Export

Updated September 3, 2026

The order that avoids rework

1. ImportPaste or upload the manuscript; headings become chapters, blank lines become paragraphs. Fix structure now — merging paragraphs later means re-generating them.
2. CastAssign the narrator, then character voices for recurring speakers. Assignments persist through the project.
3. Pronunciation rulesEnter names and terms before generating anything: an alias respelling works on every voice tier; IPA is available for Leap Turbo voices and pinyin for Leap Natural voices.
4. Generate one chapterNot the whole book. Listen, adjust the director's note or the voice, and only then batch the rest.
5. Review and re-takeListen at 1.25× with the text open. Regenerate failing paragraphs; earlier takes stay in version history.
6. ExportPer chapter or whole book. Standard preset for streaming (-14 LUFS, 48 kHz); Audiobook preset for retailer specs (44.1 kHz, -20 LUFS mastering start, then RMS/peak/bitrate/duration checks on the final MP3). Subtitles and an optional music bed export with it.

Chapter conventions that pay off

• Name chapters exactly as the retailer will list them; export filenames follow.
• Keep front matter as its own chapter so the opening credits export as one file.
• Split any chapter over ~100 minutes; the distribution checks flag files over 120.
• Put recurring asides (recaps, epigraphs) in their own paragraphs so a change is one re-take.

Reading the distribution report

With the Audiobook preset, each export is measured after encoding: RMS between −23 and −18 dBFS, peak at or below −3 dB, 44.1 kHz, at least 192 kbps CBR, file under 120 minutes; noise floor is reported as an estimate only. A pass means the file meets the common technical spec. It does not mean a given retailer accepts AI narration — the report says so — so check policy before you upload.

Narrators to audition

Frequently asked questions