You spent weeks recording. You edited out the breaths, re-recorded the stumbles, uploaded every chapter to ACX and clicked submit. Then the email arrived: your audiobook has been returned to production.
If your ACX audiobook was rejected, you're in good company. Rejections are common, especially on a first title, and most of them come down to a short list of fixable problems. The challenge is that the reasons can be vague, the specs use audio engineering terms like RMS and noise floor, and it's not always clear what to fix first.
This guide sorts ACX rejections into three buckets, explains the official requirements in plain English, shows how to check your files with free command-line tools, and gives you a fix-it order that gets most titles approved on the next try. Every rule here is taken from ACX's own help pages.
First, Find the Rejection Email
Start by finding out exactly what ACX flagged. According to ACX's list of common quality assurance issues, a rejected title is returned to production, and ACX may email the details of the required changes. Those emails often land in spam or junk folders, so check there first.
Read the notice like a repair order. Write down every issue it mentions before you open your audio editor. Fixing one problem while missing another simply leads to a second rejection.
Bucket 1: Missing Information Rejections
These are the quickest fixes, and ACX lists them as the most common reasons a title can't be approved. None of them involve your audio at all.
- Missing copyright information. The copyright owner, copyright years and name must be entered before a title can go on sale. ACX says you can ask for them to be added through Edit Retail Information.
- Missing section titles. Every uploaded file needs a chapter or section name. These names are internal only and aren't shown to listeners, but ACX uses them to spot files that are out of order.
- Missing cover art. Every title needs cover art that meets ACX's cover requirements, including the square format audiobook stores use.
- Missing retail sample. The retail sample is the short clip shoppers hear before buying. It must be five minutes or less, uploaded in the Retail Audio Sample section.
Fix all of these first. They take minutes, and getting them out of the way means any remaining rejection is about the audio itself.
Bucket 2: Technical Audio Rejections
The Specs at a Glance
The ACX audio submission requirements set out these technical rules:
- Format: each file must be a 192 kbps or higher constant bit rate (CBR) MP3 at 44.1 kHz.
- Volume (RMS): each file must fall between −23 dB and −18 dB RMS.
- Peaks: no peak higher than −3 dB.
- Noise floor: no higher than −60 dB RMS.
- Room tone: 1 to 5 seconds of room tone at the head and tail of each file, and never more than 5 seconds.
- Channels: all files must be in the same channel format, either all mono or all stereo.
- Length: no file longer than 120 minutes.
Here's what the most common technical failures mean and how to fix them.
RMS Too Low or Too High
RMS is a measure of average loudness. If a file is below −23 dB, listeners have to turn the volume up. If it's above −18 dB, it's tiring to listen to and often over-compressed.
Quiet files usually come from soft delivery or a microphone placed too far away. Loud files usually come from aggressive mastering. Fix it with sensible gain staging when recording, then bring each finished chapter into range with normalization, a gentle compressor or a loudness tool, measuring again afterwards.
Peaks Over −3 dB
Peaks are the loudest instantaneous moments: a plosive "p," a laugh, a shouted line of dialogue. ACX wants every peak at −3 dB or lower to prevent distortion. A limiter set just below −3 dB catches most stray peaks. If a line is truly clipped and distorted in the original recording, re-record it, because no limiter can repair clipping.
Noise Floor Above −60 dB
The noise floor is the level of background sound when nobody is speaking: computer fans, air conditioning, traffic, the hum of a fridge. ACX's limit of −60 dB is strict, and it's the most common technical reason home recordings fail.
Fix it at the source first. Turn off fans and appliances, record in a smaller, softer space, move the microphone closer to your mouth and away from your computer. Only then use noise reduction, and use it gently. Heavy noise reduction creates watery, metallic artifacts that a human reviewer can still reject, even if the numbers pass. Narrators on the Audacity forum have reported exactly this: files that met the measurements but were rejected anyway because the processing was audible.
Wrong MP3 Settings and Room Tone
Other common technical problems include:
- Exporting variable bit rate (VBR) MP3s instead of constant bit rate
- Exporting at 48 kHz instead of 44.1 kHz
- Mixing mono and stereo files in one project
- Missing room tone at the start or end of files, or leaving long silences of more than five seconds
These are export settings, so they're quick to fix once you know what to look for.
Check and Fix Files With ffmpeg
Free tools can measure all of this before you upload. ACX offers its own ACX audio analysis tool, and many narrators use Audacity's ACX Check plugin. If you're comfortable with the command line, ffmpeg and ffprobe can check a whole folder of chapters at once.
Export Masters to ACX's Format
Start from a high-quality master file, such as WAV, for each chapter. This command converts it to a 192 kbps CBR, 44.1 kHz, mono MP3:
ffmpeg -i ch01-master.wav -ac 1 -ar 44100 -codec:a libmp3lame -b:a 192k ch01.mp3
Setting a fixed bit rate with -b:a makes the MP3 encoder produce constant bit rate files. Use -ac 2 instead of -ac 1 only if your whole project is in stereo.
Measure Every File
This script checks the format, overall RMS, peak level and the RMS of the first second, which should be your room tone, for every MP3 in the folder:
for f in *.mp3; do
echo "== $f"
ffprobe -v error -select_streams a:0 \
-show_entries stream=codec_name,sample_rate,channels,bit_rate \
-of default=noprint_wrappers=1 "$f"
ffmpeg -hide_banner -nostats -i "$f" \
-af astats=metadata=0:measure_perchannel=none:measure_overall=RMS_level+Peak_level \
-f null - 2>&1 | grep -E "RMS level|Peak level"
ffmpeg -hide_banner -nostats -t 1 -i "$f" \
-af astats=metadata=0:measure_perchannel=none:measure_overall=RMS_level \
-f null - 2>&1 | grep "RMS level" | sed 's/RMS level/Room tone (first 1s) RMS/'
done
For each file, you want to see:
-
codec_name=mp3,sample_rate=44100, the right number of channels andbit_rate=192000or higher - An overall RMS level between −23 and −18 dB
- A peak level of −3 dB or lower
- A room tone RMS of −60 dB or lower, which is a good stand-in for the noise floor because it measures the silence before you start speaking
Treat these numbers as a pre-flight check, not a guarantee. Tools measure levels slightly differently, and ACX's quality reviewers also listen for things no meter catches, such as audible edits, mouth clicks, mispronunciations and inconsistent sound between chapters.
Bucket 3: Structure, Credits and Eligibility
One Chapter or Section Per File
ACX requires each file to contain only one chapter or section. Don't merge several short chapters into one file, and don't split a chapter across files unless it's longer than the 120-minute limit. Name your files and section titles consistently so the order is obvious.
Opening and Closing Credits
ACX says opening credits must state the audiobook's title, author or authors, and narrator or narrators. Closing credits must signal that the book has finished, and ACX suggests ending with "The End." Missing or incomplete credits are a common, easily avoided rejection.
The Retail Sample
Your retail sample must be five minutes or less, and ACX advises using the first five minutes of the audiobook. Think of it as your best narration sample. It's what shoppers hear before deciding to buy, so it should showcase clear, engaging narration rather than credits or a slow opening.
Consistency Across Chapters
ACX expects consistent sound throughout the book, including audio levels, tone, noise level, spacing and pronunciation. A chapter recorded months later in a different room, or a character name pronounced two ways, can trigger a rejection. Keep notes on microphone position, settings and pronunciations from the first session, and listen to the transitions between chapters before uploading.
AI and Text-to-Speech Narration
This is the eligibility rule that catches many new authors. ACX's requirements state that your audiobook "must be narrated by a human unless otherwise authorized," and that unauthorized use of text-to-speech, AI or automated recordings is prohibited.
Amazon does run its own authorized synthetic narration programs, but that isn't the same as uploading AI-generated files from a third-party tool to ACX. If you're considering AI narration, check ACX's current rules for your exact situation before you produce anything, because the policy may change over time.
Fix in This Order, Then Resubmit
When you get a rejection, work through the problems in this order:
- Missing information: copyright details, section titles, cover art and retail sample.
- File structure: one chapter per file, correct names, nothing over 120 minutes.
- Credits and sample: complete opening and closing credits, and a sample of five minutes or less.
- Technical levels: format, RMS, peaks, noise floor and room tone, measured on every file.
- A full listen: check every chapter against the manuscript for mistakes, edits and pronunciation.
- Re-upload and resubmit: replace only the files you changed, then submit again.
Keep a simple quality log for each chapter, recording its RMS, peak and noise floor values and the date it was checked. It turns the next resubmission, and your next audiobook, into a quick routine.
DIY or Professional Production?
Producing an audiobook yourself is very achievable if you have a quiet, treated space, the patience to edit carefully, and someone to proof the recording against the manuscript. Many authors narrate their own nonfiction successfully.
If you'd rather focus on writing, professional production covers casting, direction, editing, word-for-word proofing, pickups and mastering. Paladin Press offers audiobook production services with human narrators, mastering to ACX specs and checks on levels and file structure before upload. No producer can promise approval, but a careful process removes the most common reasons for rejection before the files ever reach ACX.
Frequently Asked Questions
Why was my audiobook rejected by ACX?
The most common reasons are missing information, such as copyright details, section titles, cover art or a retail sample, and technical problems with RMS, peaks, noise floor, file format or room tone. Structure issues, missing credits, inconsistent sound and unauthorized AI narration can also cause rejections. Check your email, including spam, for ACX's notes.
What are ACX's audio requirements?
Each file must be a 192 kbps or higher constant bit rate MP3 at 44.1 kHz, with RMS between −23 dB and −18 dB, peaks no higher than −3 dB and a noise floor no higher than −60 dB RMS. Files need 1 to 5 seconds of room tone at each end, must all be mono or all stereo, and can't exceed 120 minutes.
How long does ACX review take after resubmission?
Review times vary with ACX's workload and can range from days to several weeks. Resubmitting resets your place in the review process, so fix every listed issue before submitting again. Check ACX's help center or your title's status page for current guidance, and watch your spam folder for updates.
Can I use AI narration on ACX?
ACX requires audiobooks to be narrated by a human unless otherwise authorized, and it prohibits unauthorized text-to-speech, AI or automated recordings. Amazon runs its own authorized synthetic narration programs, but that's different from uploading AI files from another tool. Check the current rules before producing an AI-narrated audiobook.
Does passing the ACX audio check guarantee approval?
No. Meters and analysis tools confirm technical levels, but ACX's quality review also includes human listening. Reviewers can reject files for audible edits, heavy noise reduction artifacts, mispronunciations, missing credits or inconsistent sound between chapters, even when every measurement passes.
What is room tone and how much does ACX need?
Room tone is the natural sound of your recording space with nobody speaking. ACX asks for 1 to 5 seconds of room tone at the start and end of each file, and says spacing must never exceed 5 seconds. Measuring the level of that silence is also a good way to check your noise floor.
Wrapping Up
Most ACX rejections come from a predictable list: missing information, levels outside the specs, files structured incorrectly, incomplete credits or narration that isn't authorized. Find the rejection email, fix the problems in order, measure every file, listen to the whole book once more, and resubmit with confidence.
If you'd like experienced help, consider human-narrated audiobooks mastered to ACX specs from a production team, or explore Paladin's wider book publishing services for every format of your book.
Top comments (0)