Why did ACX reject my audiobook files?

ACX usually rejects or sends audiobook files back for revision for one of six reasons (could be much more!): the audio does not meet technical specs, the file structure is wrong, the credits are missing or inconsistent, the retail sample is invalid, the audio does not match the book listing, or the narration method is not allowed for the submission path being used.

If your audiobook files were rejected, do not start by regenerating or rerecording the whole book. Start by finding the exact failure type.

A loudness failure needs a different fix from a metadata mismatch. A missing opening credit is not the same problem as a high noise floor. An AI narration eligibility issue is not solved by better mastering. Treat the rejection like a diagnosis, not a personal failure.

Important note: no tool, checklist, export setting, or production workflow can guarantee ACX approval. ACX, Audible, Amazon, and related programs make their own review decisions based on current eligibility rules, rights, content requirements, technical standards, and quality review. The goal is to prepare cleaner, better-organized audiobook files and reduce preventable production issues before review.

First, separate technical rejection from eligibility rejection

Before you fix the audio, check what kind of rejection you received.

There are two broad categories.

Technical or production issues:

• RMS volume is outside the accepted range
• Peak level is too high
• Noise floor is too high
• MP3 settings are wrong
• Room tone is too long, too short, or inconsistent
• Files are too long
• Chapters are combined incorrectly
• Opening or closing credits are missing
• Retail sample is wrong
• Metadata does not match the spoken audio
• Audio quality is inconsistent across chapters

Eligibility, rights, or content issues:

• The audiobook rights are unclear
• The narration method is not allowed for that submission path
• The voice used is not properly authorized
• The narrator label or disclosure is inaccurate
• The content does not match the submitted book
• The production route requires authorization

This distinction matters most for AI-narrated audiobooks , voice-cloned audiobooks, and text-to-speech audiobook files.

ACX’s audio submission requirements state 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 also have authorized synthetic narration programs, including Virtual Voice and Voice Replica. That does not mean every third-party AI audiobook can be submitted through every ACX path.

So if ACX rejected your AI audiobook files, check eligibility first. A technically clean file can still run into problems if the submission route does not allow that narration method.

Official reference: ACX audio submission requirements

Read the rejection notice like a repair order

Do not skim the rejection email and guess.

Look for:

• the exact file name mentioned
• the chapter or section mentioned
• any timestamps
• the reason category
• whether the issue appears in one file or every file
• whether ACX mentions audio, credits, metadata, retail sample, or narration method

Then group the problem.

If one file is rejected, the issue may be local: a bad export, missing header, wrong silence, or chapter-specific noise.

If every file is rejected, the issue is probably global: wrong MP3 settings, incorrect loudness, mixed mono/stereo files, no room tone, or a production-wide mastering issue.

If the message mentions narration method, rights, or authorization, do not try to solve it with an audio editor. Confirm the publishing path and current platform rules first.

ACX technical specs that commonly cause rejection

ACX has specific technical requirements for audiobook audio files.

Use these targets when checking your files:

These are not cosmetic settings. They affect whether the audiobook can be processed cleanly and whether listeners get consistent playback volume.

The three numbers that cause the most confusion are RMS, peak, and noise floor.

RMS rejection means the audiobook is too quiet or too loud on average

RMS measures average loudness.

ACX requires each file to sit between -23 dB and -18 dB RMS. If the file is below that range, the audiobook may sound too quiet. If it is above that range, the audiobook may sound too loud or overprocessed.

Common causes:

• unmastered narration
• inconsistent chapter volume
• aggressive compression
• exported files from different sessions
• regenerated chapters with different loudness settings
• music or effects pushing the average level too high

How to fix it:

1. Measure RMS for every file, not only the first chapter.
2. Normalize or master the files consistently.
3. Re-export with the same settings across all chapters.
4. Recheck the loudest and quietest chapters.
5. Listen after fixing numbers, because technically correct audio can still sound uneven.

Do not fix RMS chapter by chapter without checking the full book. The listener should not need to adjust volume every time a new chapter starts.

Peak rejection means the audio gets too loud at its loudest moments

Peak level measures the loudest point in the file.

ACX requires peaks no higher than -3 dB. If peaks exceed that, the file can distort or create encoding problems.

Common causes:

• clipping
• loud character dialogue
• music in credits
• harsh consonants
• sudden emphasis
• bad limiting
• exported audio with no headroom

How to fix it:

• lower peak levels below -3 dB
• use a limiter carefully
• check loud dialogue and chapter openings
• make sure credit music does not hide or overpower narration
• re-export and measure again

Do not simply reduce the whole file without checking RMS afterward. Lowering peak level can change the relationship between peak and average loudness.

Noise floor rejection means the quiet parts are not clean enough

Noise floor measures background noise.

ACX requires the noise floor to be no higher than -60 dB RMS. In human-recorded audio, this often means hiss, room hum, fan noise, mouth clicks, microphone noise, or room echo. In AI-narrated audiobook files, the problem may be different: metallic artifacts, glitches, regenerated sections that do not match, or noise introduced during editing.

Common causes:

• noisy recording environment
• overused noise reduction
• low-quality exports
• background hum
• mouth sounds
• room echo
• audio glitches
• generated sections pasted beside older sections

How to fix it:

1. Identify whether the noise is throughout the file or only in specific sections.
2. Remove obvious clicks, pops, and glitches.
3. Re-export weak AI-generated passages instead of overprocessing them.
4. Check quiet passages and chapter endings.
5. Listen on headphones and speakers.

Noise floor is not just a number. If the listener hears distractions during quiet passages, the audiobook still needs work.

Wrong MP3 settings can trigger avoidable rejection

ACX requires 192 kbps or higher Constant Bit Rate MP3 files at 44.1 kHz.

Avoid:

• Variable Bit Rate MP3
• low-bitrate MP3
• WAV uploaded in place of MP3
• M4A
• inconsistent sample rates
• mixed mono and stereo files
• files exported from different tools with different settings

If your audiobook was rejected for formatting, check every file in the folder. One bad file can delay the whole submission.

Good export habit:

• export all final upload files in one controlled session
• use the same bit rate, sample rate, and channel format
• keep a separate archive of higher-quality source files if needed
• name final upload files clearly

Narration Box can help keep audiobook chapters organized and exported consistently, but you should still verify the final files against the current requirements before uploading.

Room tone and silence problems are easy to miss

ACX recommends room tone at the beginning and end of each file, generally between 1 and 5 seconds.

Too little space can make a file feel abrupt. Too much silence can feel broken. Inconsistent spacing across chapters makes the audiobook feel poorly assembled.

Check:

• the file does not start with a clipped first word
• the opening room tone is not too long
• the ending does not cut off the last word
• music does not hide the final narration
• each chapter has a similar spacing pattern
• there are no long accidental silences after headings

For AI-narrated audiobooks , also check generated pauses. A pause after a chapter title can help. A long silence inside a sentence can make the narration feel broken.

Chapter structure rejection means the audiobook is assembled wrong

ACX expects one chapter or section per file. Opening and closing credits should be separate files.

A rejected audiobook may have perfectly good audio but poor structure.

Risky structure:

• one giant MP3 for the whole book
• several chapters inside one file
• missing opening credits file
• missing closing credits file
• a chapter longer than 120 minutes
• files named vaguely, such as final_audio.mp3
• file names with special characters
• chapter headers missing inside the audio

Better structure:

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

Each file should contain one clear section. If a chapter is longer than 120 minutes, split it into separate files and make the continuation clear to the listener.

Narration Box’s audiobook creator can help here because long books can be prepared chapter by chapter instead of as one unreviewed audio block.

Missing opening or closing credits can delay review

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

A clean opening credit usually includes:

• book title
• author name
• narrator name or approved narration label
• edition wording, if needed

Example:

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

A clean closing credit may sound like:

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

The exact wording should match the book metadata and the platform’s current requirements. If your narration method requires a specific label or disclosure, follow the rule for that submission path.

Avoid:

• spoken title differs from listing title
• author name pronounced differently from metadata
• narrator name differs between audio and listing
• closing credit does not indicate the book has ended
• music covers the credit narration
• AI narration is labeled as a human narrator when disclosure is required

Credits are small files, but they carry a lot of review risk.

Retail sample rejection means the sample is not usable

ACX requires a retail sample of 5 minutes or less. The sample must come from the audiobook itself.

Do not submit:

• a trailer
• a sales pitch
• a separate teaser
• a rewritten summary
• a sample longer than 5 minutes
• a sample with explicit content when a cleaner section is possible
• audio that does not match the final narration quality

A good retail sample should show the real listener experience.

For fiction, choose a passage with narration quality, scene flow, and clean character delivery.

For nonfiction, choose a section that shows clarity, pacing, and authority.

For educational or technical books, choose a section where terms and examples are handled cleanly.

If the first minutes of the book are mostly front matter, dedication, legal text, or explicit content, check whether your submission path allows a more suitable excerpt.

Metadata mismatch rejection means the audio and listing disagree

ACX reviews whether the audiobook audio matches the book listing.

Check:

• title
• subtitle
• series name
• volume number
• author name
• narrator name or narration label
• publisher name
• language
• cover text
• chapter titles
• opening credit wording
• closing credit wording
• retail sample source

Common mismatch examples:

• the audio says a shorter title than the listing
• the subtitle appears on the cover but not in credits
• the author uses initials in one place and full name in another
• narrator name differs between intro and metadata
• chapter names do not match the file structure
• the retail sample uses a different version of the audio

Metadata problems are frustrating because the audio may sound fine. Fix them before making any assumptions about mastering.

Manuscript mismatch means the narration does not match the book

The audiobook should match the submitted or referenced book text closely.

This is especially important when converting PDF, DOCX, EPUB, or manuscript files into audio. Import problems can create hidden mistakes.

Look for:

• missing chapters
• duplicated paragraphs
• skipped footnotes
• broken dialogue
• repeated PDF headers and footers
• page numbers read aloud
• hyphenated words split incorrectly
• tables or figures removed without explanation
• old manuscript draft used by mistake
• AI pronunciation errors that change meaning

If you are using an AI audiobook generator or AI audiobook creator, clean the manuscript before generating audio. Remove running headers, footers, page numbers, broken line breaks, and formatting artifacts.

For long books, work chapter by chapter. It is easier to find a bad paragraph in Chapter 7 than inside one huge generated file.

AI narration rejection requires a different fix

AI narration creates two separate review questions.

1. Is the audiobook eligible under the submission path?
2. Are the audio files technically and editorially prepared well?

Do not confuse those two.

Technical preparation can help with file quality, structure, pronunciation, consistency, and listener experience. It cannot override ACX, Audible, Amazon, KDP, publisher, or rights-holder rules.

Before submitting an AI-narrated audiobook, confirm:

• the exact program or platform allows that narration method
• the voice is licensed or authorized
• any cloned voice has consent
• the narrator label is accurate
• disclosure follows current platform requirements
• you own or control the audiobook rights
• the audio files meet technical specs

If you are unsure whether your audiobook is eligible, check current ACX and Audible documentation before producing or uploading the full audiobook.

Fix the rejection in the right order

When authors panic, they often fix the wrong thing first.

Use this order:

1. Read the rejection notice.
2. Identify whether the problem is eligibility, technical audio, structure, credits, retail sample, metadata, or manuscript match.
3. Fix eligibility or rights issues before touching audio.
4. Fix global export settings before fixing individual chapters.
5. Fix chapter structure and credits before selecting the retail sample.
6. Fix pronunciation and pacing before the final listen.
7. Re-export only the affected files if possible.
8. Measure the corrected files again.
9. Listen to the corrected sections beside surrounding audio.
10. Keep a clean final folder for upload.

This order saves time because it prevents you from mastering files that later need to be regenerated.

How Narration Box helps with audiobook repair and prevention

Narration Box is useful when the issue is production control, not platform approval.

It can help authors and production teams:

• organize audiobook narration by chapter
• edit manuscript text before generating audio
• use consistent narrators across chapters
• add style instructions for pacing and tone
• fix pronunciation for names, terms, acronyms, and non-English words
• regenerate selected paragraphs or sections after edits
• export chapter-level audio files
• keep narration more consistent across long-form projects

That does not mean Narration Box guarantees ACX approval. It does not. Final acceptance depends on the platform’s current rules and review process.

A practical repair workflow:

1. Import or paste the final manuscript.
2. Split the project into chapters and credits.
3. Confirm the narrator or voice is allowed for your publishing route.
4. Generate one corrected chapter first.
5. Listen for pacing, pronunciation, and artifacts.
6. Regenerate weak paragraphs instead of rebuilding the whole book.
7. Export chapter files consistently.
8. Check audio specs before upload.
9. Verify metadata and credits against the book listing.
10. Keep the final files organized for submission.

The main advantage is control. If Chapter 9 has a mispronounced character name, you should not have to rebuild the entire audiobook. Fix the affected section, regenerate it, and recheck the join.

What to check before resubmitting

Before resubmitting rejected audiobook files, confirm:

• the rejection reason has been fixed
• the audiobook is eligible for the submission path
• every file is MP3
• every file is 192 kbps or higher CBR
• every file is 44.1 kHz
• RMS is between -23 dB and -18 dB
• peaks stay below -3 dB
• noise floor stays below -60 dB RMS
• mono/stereo format is consistent
• every file is under 120 minutes
• opening credits are included
• closing credits are included
• retail sample is 5 minutes or less
• sample comes from the audiobook
• chapter files contain one chapter or section each
• file names avoid special characters
• spoken credits match metadata
• narration matches the manuscript
• pronunciation is checked across the whole book
• regenerated sections sound natural beside unchanged sections
• the full audiobook has been listened to from start to finish

The full listen is not optional. Some of the most expensive problems are hidden in the middle of the book.

When to regenerate, remaster, or restructure

Different problems need different fixes.

Do not remaster when the issue is metadata. Do not rewrite the script when the issue is peak level. Do not re-export files before confirming eligibility.

Final recommendation

If ACX rejected your audiobook files, slow down and diagnose the exact reason.

Most production problems fall into a small set of categories: loudness, peaks, noise floor, room tone, MP3 settings, chapter structure, credits, retail sample, metadata, manuscript match, pronunciation, or consistency.

AI-narrated audiobooks add one more layer: eligibility. Check whether your narration method is allowed under the exact ACX, Audible, KDP, publisher, or distribution path you are using.

Once the route is clear, focus on the listener. The goal is not only to satisfy waveform numbers. The goal is to produce a clean, consistent audiobook that is easy to follow from the first file to the last.

Narration Box can help with chapter-based audiobook production, pronunciation fixes, style direction, block-level edits, and consistent exports. It is a preparation and production tool, not an approval guarantee.

Note: Narration Box helps authors prepare audiobook files for submission, but eligibility, review, approval, and rejection decisions are made solely by ACX, Audible, Amazon, or the relevant distribution platform under their current policies.