ACX Rejection Checklist for AI-Narrated Audiobooks

ACX rejection usually comes from one of three problems: the audiobook is not eligible for the submission path being used, the files fail technical audio requirements, or the audiobook structure does not match ACX formatting rules.

For AI-narrated audiobooks , the first check matters most. ACX currently states that submitted audiobooks must be narrated by a human unless otherwise authorized, and that unauthorized text-to-speech, AI, or automated recordings are prohibited. Audible and Amazon do support synthetic narration through specific programs such as Virtual Voice and Voice Replica, but that does not mean every third-party AI audiobook can be uploaded through every ACX path without authorization.

Use this checklist before submitting an AI audiobook , voice-cloned audiobook, or text-to-speech audiobook for Audible, Amazon, and Apple distribution.

Fast Pre-Upload Verdict

Before you upload, confirm these four things.

• Your audiobook is allowed under the exact ACX, Audible, KDP, or publisher program you are using.
• Every chapter is exported as its own MP3 file, not one giant audiobook file.
• Every file meets ACX technical specs: 192 kbps or higher CBR MP3, 44.1 kHz, -23 dB to -18 dB RMS, peaks below -3 dB, and noise floor below -60 dB RMS.
• The spoken opening credits, closing credits, retail sample, chapter headers, file names, and metadata all match the book listing.

If any one of these fails, your audiobook can be delayed, sent back for revision, or rejected.

ACX AI Audiobook Eligibility Check

AI narration must be checked against the submission route before audio quality is checked.

ACX’s audio submission requirements say that submitted audiobooks must be human narrated unless otherwise authorized. This creates a practical distinction:

• A human-narrated audiobook submitted through a standard ACX workflow follows normal ACX rules.
• A Virtual Voice audiobook created inside Amazon or Audible’s authorized systems follows that program’s rules.
• A Voice Replica audiobook follows Audible’s specific voice replica agreement and eligibility rules.
• A third-party AI-narrated audiobook may require authorization before ACX submission.

This is the part many authors miss. A technically perfect AI audiobook can still fail if the submission path does not allow that narration method. You can use our free ACX compliance checker tool for free to roughly check if your manuscript is aligned with ACX requirements.

Before producing the full audiobook, check the current ACX and Audible policy pages, confirm your account’s eligibility, and keep a record of the program or authorization route you are using.

Useful official references:

• ACX audio submission requirements
• KDP audiobooks with Virtual Voice
• Audible Virtual Voice and Voice Replica

Rights and Voice Consent Check

Voice rights can block an audiobook even when the audio sounds professional.

For AI audiobook narration, confirm the following before export:

• You own the audiobook rights for the book.
• You own or have licensed the voice used in the audiobook.
• Any cloned voice was created from your own voice or an approved voice sample.
• You have written permission if the voice belongs to another narrator, author, trainer, actor, or public figure.
• The narrator name, AI disclosure, and production method are accurate for the platform you are using.

Do not use a celebrity-like voice, a narrator imitation, a synthetic voice trained from an unapproved sample, or a clone of another person without written consent. Voice cloning can be useful for authors and publishers, but only when the voice is legally cleared.

If you are using Narration Box, keep your voice cloning workflow permission-aware: use your own voice or an approved voice sample, then save the final narrator selection for consistent audiobook production.

ACX Technical Audio Specs

ACX checks audiobook files against strict audio standards.

Each uploaded file should meet these requirements:

These specs are not optional. A file that sounds fine in headphones can still fail because its RMS is too low, the peak level clips, the noise floor is too high, or the MP3 was exported as variable bit rate.

Before upload, run each file through an audio checker. You can use the Narration Box ACX Compliance Checker to catch common technical issues before submission.

Chapter File Structure Check

ACX expects one chapter or section per file.

Your audiobook should not be uploaded as one long MP3. It should be split into clean, labeled files:

• Opening credits
• Chapter 1
• Chapter 2
• Chapter 3
• Each remaining chapter or section
• Closing credits

Each file should contain only one chapter or section. If a chapter runs longer than 120 minutes, split it into separate files and add a clear secondary header so the listener understands the continuation.

Good file structure:

• 00_opening_credits.mp3
• 01_prologue.mp3
• 02_chapter_1.mp3
• 03_chapter_2.mp3
• 20_closing_credits.mp3

Risky file structure:

• my_audiobook_final.mp3
• part1.mp3
• audio_export_new_latest_fixed.mp3
• chapter_1_2_3.mp3

Narration Box’s block-based audiobook workflow helps here because chapters, sections, author notes, and credits can be generated and exported separately instead of forcing the whole book into one audio file.

Opening Credits Check

Opening credits must identify the audiobook clearly.

The spoken opening credit should include:

• Book title
• Author name
• Narrator name or approved narration label
• Any required edition wording if relevant

A clean opening credit sounds like this:

This is The Midnight Orchard, written by Maya Rivers, narrated by Sarah Lane.

For AI narration, use the narrator label required by your distribution path. Do not invent a misleading human narrator name if the platform requires AI narration disclosure.

The opening credit should match the book metadata, cover, title listing, and narrator field. Small mismatches can create avoidable review delays.

Closing Credits Check

Closing credits must clearly tell the listener the audiobook is complete.

A clean closing credit sounds like this:

You have been listening to The Midnight Orchard, written by Maya Rivers, narrated by Sarah Lane. The End.

The closing credit should not trail off awkwardly, cut suddenly, or end with music that hides the final words. ACX allows music in credits, but the narration must remain clear.

For AI audiobook narration, use the same narrator label and voice style used in the opening credits unless the platform requires a different disclosure format.

Retail Sample Check

The retail sample is a sales asset and a compliance file.

ACX requires a retail sample of 5 minutes or less. The sample must come from the audiobook itself. It should not be a trailer, an ad, a separate AI-generated teaser, or a rewritten summary.

A good retail sample should:

• Come directly from the final audiobook
• Be 5 minutes or less
• Avoid explicit content when possible
• Start cleanly without confusing context
• Show the real narration quality
• Match the final voice, pacing, and mastering

For fiction, choose a sample with narration quality, character flow, and emotional control. For nonfiction, choose a section that quickly shows authority, clarity, and pacing. For educational books, choose a section where the AI narrator handles terms, examples, and transitions cleanly.

If the first five minutes are mostly front matter, legal text, dedication, or explicit content, select a better sample if your submission path allows it.

Metadata Match Check

ACX reviews the relationship between the audio and the book listing.

Check every metadata field before upload:

• Title
• Subtitle
• Series name
• Volume number
• Author name
• Narrator name
• Publisher name
• Rights holder
• Language
• Cover text
• Chapter names
• Opening credit wording
• Closing credit wording

The audiobook should not say one title while the listing shows another. The narrator credit should not differ between the spoken intro, metadata, and cover references. If the ebook title includes a subtitle, decide whether the audio credits should include it and keep that choice consistent.

Metadata mistakes are frustrating because they are easy to prevent and can delay a release even when the audio files are technically clean.

Manuscript Match Check

The narrated audiobook should match the published manuscript closely.

ACX can flag issues when the audio does not match the book’s content. This is especially important when using an AI audiobook generator or AI audiobook creator because it is easy to paste an outdated manuscript, remove headings, skip front matter, or accidentally generate from a draft.

Before final export, compare the audio project against the book version being submitted.

Check for:

• Missing chapters
• Duplicate paragraphs
• Skipped footnotes or endnotes
• Repeated headers and footers from PDFs
• Broken dialogue
• Incorrect section order
• Missing author notes
• Deleted tables or figures that need spoken explanation
• AI pronunciation errors that change meaning

If you are converting a PDF, DOCX, or EPUB into an audiobook, clean the manuscript first. Remove page numbers, running headers, broken hyphenation, and repeated footer text before generating narration.

For long-form books, use the Narration Box audiobook creator to work chapter by chapter instead of generating from one unreviewed text block.

Pronunciation QA Check

Pronunciation errors hurt listener trust and can trigger revision work.

AI narration is strongest when names, places, acronyms, technical terms, fantasy words, and brand names are handled before final export.

Create a pronunciation list for:

• Character names
• Place names
• Author name
• Series title
• Fictional terms
• Medical terms
• Legal terms
• Academic terms
• Acronyms
• Non-English words
• Brand names
• Historical names

Then listen to every occurrence in the audiobook. Do not only test the first chapter. A name can sound correct in one context and wrong in another if the surrounding text changes.

In Narration Box, use pronunciation control and block-level regeneration to fix only the affected paragraph or chapter instead of recreating the full audiobook.

Pacing and Pause Check

AI audiobook narration must sound consistent across long listening sessions.

ACX cares about listener experience, not just waveform numbers. A file can pass technical specs and still feel poor if the narration is rushed, flat, uneven, or full of awkward pauses.

Check pacing at three levels:

• Sentence pacing: Each sentence should be easy to follow without sounding slow.
• Paragraph pacing: Scene changes, argument shifts, and section transitions need breathing room.
• Chapter pacing: The book should feel consistent across the full listening experience.

Common AI narration problems include:

• No pause after chapter titles
• Rushed dialogue
• Overly dramatic emotional tags
• Flat nonfiction delivery
• Inconsistent speed between chapters
• Long silences after headings
• Pauses that make sentences feel broken
• Unnatural emphasis on ordinary words

Use style instructions carefully. A prompt like “calm, clear, controlled audiobook narration with natural pauses” is usually better than trying to force emotion into every paragraph.

Consistency Check Across Chapters

ACX expects consistent sound, formatting, tone, spacing, and pronunciation.

For AI-narrated audiobooks, consistency is one of the main advantages if the workflow is controlled. It is also a common failure point if the book is generated in scattered sessions with different voices, different settings, or different prompts.

Before submission, confirm that every chapter uses the same:

• Narrator or approved narrator plan
• Voice model
• Language and accent
• Speaking speed
• Style instruction
• Pronunciation rules
• Export format
• Loudness target
• Room tone pattern
• Chapter header format

If you intentionally use multiple voices, document the plan. For example, a nonfiction book may use one narrator for the main text and a different approved voice for quoted interviews. A fiction audiobook may use narrator blocks and dialogue blocks. Keep that structure clear, consistent, and legally cleared.

Noise and Artifact Check

AI narration should be checked for both audio noise and generation artifacts.

Traditional audiobook noise includes:

• Hiss
• Hum
• Clicks
• Pops
• Mouth noise
• Room echo
• Breath blasts
• Background sound
• Edits that cut too sharply

AI narration can add different issues:

• Warbled words
• Metallic consonants
• Sudden tone changes
• Misread punctuation
• Over-smoothed emotion
• Repeated phrase rhythm
• Robotic sentence endings
• Strange pronunciation on names
• Audio glitches around regenerated sections

Listen to the audiobook like a buyer, not only like a producer. Use headphones and speakers. Check quiet chapters and loud chapters. Check regenerated blocks beside unchanged blocks so the edit does not sound pasted in.

Export Settings Check

Export settings must be uniform across the full audiobook.

Use the same export settings for every file:

• MP3
• 192 kbps or higher
• Constant Bit Rate
• 44.1 kHz
• Mono or stereo, consistent across all files
• Clean file names using standard letters and numbers
• No special characters in file names
• No accidental WAV, M4A, VBR MP3, or low-bitrate exports

If your AI audiobook creator exports multiple formats, choose the ACX-ready version for ACX upload. Use WAV or higher-quality formats for editing and archiving, but upload the format required by ACX.

Narration Box supports high-quality audiobook exports and chapter-based workflows, which makes it easier to keep files organized before the final ACX compliance check.

Final Listening Pass

A full listen is the last quality gate.

Do not submit after spot-checking only the first chapter. ACX review delays often come from problems buried deep in the audiobook: a missing chapter, a repeated paragraph, one broken export, one wrong voice, or one file with a different loudness profile.

During the final listen, track issues in a simple table:

For AI narration, block-level editing saves time. Fix the smallest affected section, regenerate that block, then re-check the join.

Upload Readiness Checklist

Use this final checklist immediately before submission.

• Confirm the audiobook is allowed under your ACX, Audible, KDP, or publisher submission path.
• Confirm audio rights and voice rights are documented.
• Confirm AI narration or voice cloning disclosure follows the platform’s current rule.
• Export each chapter or section as its own MP3 file.
• Keep every file under 120 minutes.
• Use 192 kbps or higher Constant Bit Rate MP3.
• Use 44.1 kHz sample rate.
• Keep RMS between -23 dB and -18 dB.
• Keep peak levels below -3 dB.
• Keep noise floor below -60 dB RMS.
• Use consistent mono or stereo across all files.
• Add 1 to 5 seconds of room tone at the beginning and end.
• Include opening credits as a separate file.
• Include closing credits as a separate file.
• Include a retail sample of 5 minutes or less.
• Match spoken credits to metadata.
• Match chapter headers to file structure.
• Remove special characters from file names.
• Check pronunciation for names, terms, and acronyms.
• Listen to the full audiobook from start to finish.
• Run files through an ACX audio checker before upload.

Narration Box Workflow for Lower-Risk ACX Preparation

Narration Box helps authors prepare AI audiobook narration in a controlled, chapter-based workflow.

A practical workflow looks like this:

1. Import or paste the final manuscript into Narration Box.
2. Split the audiobook into chapters, credits, author notes, and bonus sections.
3. Choose an approved AI narrator or approved cloned voice.
4. Add book-level style instructions for consistent pacing and tone.
5. Add pronunciation rules for names, acronyms, and book-specific terms.
6. Generate one chapter first and listen before producing the full book.
7. Regenerate only weak paragraphs or sections.
8. Export chapter-level files with consistent settings.
9. Check the files with the ACX Compliance Checker .
10. Verify platform eligibility and submission rules before upload.

This workflow does not replace ACX authorization or platform review. It helps reduce avoidable file, chapter, pronunciation, and export mistakes before submission.

Common ACX Rejection Triggers for AI Audiobooks

Most preventable rejection issues fall into these categories.