Audiobook Chapter Files and Metadata: A Delivery Guide

An audiobook is usually delivered as an ordered set of files, not one enormous recording. That structure affects navigation, retail samples, quality control, revisions, and what a listener sees in an app. A clean waveform cannot rescue a book whose chapters are mislabeled, out of order, or inconsistent with its metadata.

Build a manifest before export. It should be the single map between manuscript sections, production masters, delivery files, spoken headers, and platform titles.

Define the audio edition

List what the audio edition contains: opening credits, front matter, prologue, chapters, interludes, appendices, acknowledgments, endnotes, and closing credits. Decide whether visual-only material needs adaptation. Do not omit a section merely because a parser could not classify it.

Compare the list with the ebook table of contents and the approved manuscript. The W3C EPUB 3.3 specification describes the spine as the default reading order and the navigation document as the machine-readable table of contents. Those are useful inputs, but the audio edition still needs editorial review: an EPUB can contain advertisements, page-list navigation, or decorative matter that is not a spoken chapter.

Assign a stable section ID that never depends on a mutable title. “Chapter Three” may later become “The Crossing,” while ch003 continues to connect scripts, audio, defects, and revisions.

Make one manifest authoritative

Useful columns include:

• sequence number and stable section ID;
• source document or EPUB spine item;
• display title and spoken header;
• production-master filename;
• delivery filename;
• source text version or checksum;
• narrator or cast scope;
• duration;
• technical-QC status;
• editorial-QC status;
• sample eligibility;
• revision and approval date.

Do not maintain separate ordering spreadsheets for editing and upload. They will drift. Export platform-specific views from the master manifest.

Use zero-padded sequence numbers such as 001, 002, and 010. Lexical file sorting then matches intended order. Keep titles human-readable but conservative in filenames; punctuation and non-ASCII characters may be accepted by one service and rejected by another.

Align spoken and displayed titles

The listener hears the header and sees a chapter title. They should identify the same section. If the manuscript says only “3,” decide whether the audio says “Chapter Three” and record that decision.

ACX's current audio submission requirements say each file should contain one chapter or section and include its spoken header. Files longer than 120 minutes must be split. Its chapter-title guidance uses a “Continued” convention for split chapters. Recheck current distributor rules before delivery rather than assuming one platform's convention is universal.

Avoid stuffing chapter metadata with promotional language. Navigation titles should help a listener find content. Preserve capitalization consistently and remove EPUB artifacts such as fragment IDs or stray numeric labels.

Keep credits consistent

Opening credits, closing credits, cover art, and catalog metadata must agree on title, subtitle, author, and narrator. Decide how editors, translators, or multiple performers are credited based on contracts and platform fields.

Maintain approved credit text in the manifest. Do not retype it during each export. A punctuation difference may be harmless, but a missing contributor or different title is a rights and quality problem.

The closing credit should clearly signal finality. Opening and closing credits are separate files under current ACX rules. Other distributors may package credits differently, so preserve lossless masters and generate delivery sets instead of renaming the only copy.

Create a defensible retail sample

Record the sample's source file, start time, end time, and master version. The sample must come from the audiobook. Under current ACX guidance it is no longer than five minutes and should avoid explicit material.

Choose a coherent passage representative of the book's actual narration. Starting mid-sentence or ending on a clipped breath makes a poor sample even when the duration passes. If the underlying chapter changes, invalidate and regenerate the sample.

Preserve masters and derive deliveries

Keep one lossless, approved master per section. Create MP3 or other distributor formats from that master. Repeatedly editing and re-encoding an MP3 accumulates loss and makes revisions hard to audit.

A practical layout is:

masters/ch003-v003.wav
deliveries/acx/003-chapter-three.mp3
deliveries/spotify/003-chapter-three.flac
manifests/audiobook-v003.csv
reports/qc-v003.json

The filenames show which version produced a delivery. Add checksums so a file can be matched to its report after upload.

Spotify for Authors currently accepts MP3, WAV, or FLAC and asks creators to title chapters clearly. Its official upload guidance also requests a sample. Platform flexibility is another reason to preserve masters instead of treating one distributor's MP3 as the archive.

Validate order mechanically and audibly

A script can catch duplicate sequence numbers, missing files, duration limits, and mismatched checksums. It cannot determine whether the ending of Chapter Five was accidentally attached to Chapter Six.

Listen across every boundary:

• end of opening credits into the first content section;
• chapter endings into the next spoken header;
• split chapter parts;
• final content into closing credits.

Check for duplicated or missing sentences, sudden level changes, long silence, and stale headers. Boundary review is faster than replaying the entire book and finds errors isolated-file review misses.

Run the technical checks described in the ACX preflight after final encoding. Keep editorial and technical status separate: a file can pass peak and RMS limits while containing the wrong chapter.

Control revisions

Never overwrite an approved master without changing its version. A correction record should say what changed, why, who approved it, and which delivery files were regenerated.

When one section changes:

1. create a new source/script revision;
2. render or edit a new lossless master;
3. repeat editorial listening for the changed context;
4. repeat technical analysis;
5. generate every affected platform delivery;
6. update checksums and sample if applicable;
7. archive, rather than delete, the superseded version.

This prevents a fixed ACX file from coexisting with an old Spotify file under the same apparent version.

Final delivery checklist

Confirm that:

• the manifest covers every intended section exactly once;
• sequence numbers are unique and gap-free;
• spoken headers match display titles;
• credits agree with cover and catalog metadata;
• every delivery traces to an approved lossless master;
• samples trace to the current chapter version;
• split sections follow the target platform's rule;
• checksums, duration, encoding, and QC results are recorded;
• chapter boundaries passed an audible continuity review;
• the live upload order matches the manifest before submission.

VoxForge exports chaptered audio, but the producer still owns the delivery manifest and platform-specific preflight. Treat chapter metadata as part of the product, not clerical cleanup after mastering. A stable manifest makes the book navigable today and repairable when a distributor or chapter changes later.