ACX lets you add music to an audiobook. It asks you to do it "carefully and sparingly", and warns that music "can be a distraction when it overlaps with narration".[1] A music bed, a track that plays under the whole chapter, has a second problem. It fills every pause, and the pauses are where the noise floor is measured.[2] One chapter a narrator sent us to look at had a noise floor of -35.7 dBFS (decibels relative to full scale, where 0 is the loudest a file can hold) against ACX's limit of -60. Mastering cannot take music out from under a voice, so the fix comes before mastering: export the voice on its own, then master that.
Every ACX rule on this page was checked against the ACX help centre and ACX's audio glossary on 30 October 2026.
Does ACX allow background music under narration?
Yes. ACX's own article on music and sound effects permits both, with conditions. Five of its points matter here.[1]
- Use it sparingly. ACX recommends adding music "carefully and sparingly".[1]
- Watch the overlap. Music and sound effects "can be a distraction when it overlaps with narration".[1]
- Own the rights. The music must be "an original composition that you have the rights to use", and you need the rights to any sound effects too.[1]
- Whispersync can fail. Music "can also cause the Whispersync for Voice feature to fail". Whispersync links your audiobook to its ebook, and when it fails, the two versions are not linked together.[1]
- Still meet the spec. A production with music must still meet the ACX submission requirements.[1]
The last condition is the one that catches a music bed. The submission requirements apply to every file, music or not, and one of them is the noise floor.
Why does a music bed fail the ACX noise floor?
The noise floor is the level of sound in your file when you are not speaking. ACX's glossary calls it the audible level of background noise "where no narration is taking place".[2] In practice, that means your pauses: the gaps between sentences, between phrases and between words.
ACX's limit is strict: "Each file must have a noise floor no higher than -60dB RMS".[3] RMS is the average level, the way ACX measures both loudness and the noise floor.[3] ACX's glossary notes that "dB" in digital audio usually refers to dBFS.[2] The more negative the number, the quieter the sound. -60 dBFS is very quiet.
In a chapter with a music bed, there is no pause without music. Every gap carries the bed, so the floor reading measures the music, not your room. (Hiss, hum and fan noise raise the floor too. The audiobook noise floor guide covers those causes.)
Here is that chapter in detail. It runs five minutes, a narrator sent it to us to look at, and a music bed plays under the whole narration. The sound in its pauses was a sustained chord, the same notes from the first second to the last. It was not hiss. Even the quietest tenth of the chapter sat at -35 dBFS, so there was no quiet moment anywhere in it.
Mastering lowered the floor by 19.8 dB, and the chapter still missed the limit by 4.5 dB. The loudness landed where ACX wants it. The floor could not, mainly because the music sits under the voice and in every short gap between words. The chapter had also been compressed before mastering, which lifted the bed towards the level of the words. The guide on processing audio before mastering covers that side.
Can mastering take the music out from under the voice?
No. Mastering cannot separate two sounds that overlap, and in a mixed chapter the music and your voice are one sound. chapterpass lowers the noise in the pauses between phrases and never touches the sound under the voice. The page on what chapterpass does, and what it does not sets out the full scope.
That is why the chapter above improved and still failed. Lowering the pauses brought the floor down, but the bed was still there between words and under every sentence.
So the fix belongs in your editing project, where the voice and the music are still separate tracks.
How do I fix a chapter with music under the narration?
Open the project where you mixed the chapter, then work through these steps in order.
- Mute the music track. With the music muted, your pauses hold only room tone: the resting sound of your recording space, which ACX says should be "as close to silence as possible".[2]
- Switch off anything aimed at the ACX numbers. A limiter, loudness normalisation, heavy compression, heavy noise reduction or an "enhance" setting on the voice track: mastering sets the level, the peaks and the noise in the pauses itself. Tasteful EQ, a de-esser or gentle compression that keeps your pauses quiet can stay. The guide on processing audio before mastering draws the line.
- Export the voice alone as WAV. WAV is lossless, so nothing is thrown away before mastering, and the MP3 comes once, at the very end. Keep the room tone at the start and end of the take. The ACX submission checklist covers the head and tail rule.
- Check that the music is gone from the pauses. The free ACX check shows the noise floor of the voice-only file in your browser, without uploading it. Compare it with the old file's reading: the floor should have dropped. A raw voice file may still read above -60 at this stage, and that is expected before mastering.
- Master the voice-only file. Its pauses now hold room tone instead of music, which is what the noise floor is meant to measure.
- Measure the mastered file. It is the mastered MP3 that must read -60 dBFS or lower.[3] Measure it the same way. ACX's own Audio Lab gives feedback on uploaded files, but it "does not detect issues such as noise floor or editing errors".[4]
If you already cut a retail sample from the old chapter, cut it again from the new master. The retail sample is the preview of 5 minutes or less that listeners hear before they buy, and ACX says it must come from the audiobook.[3] A sample cut from the old file carries the same bed.
If the music was playing in the room while you recorded, there is no music track to mute. The music is inside the voice recording itself, and no export setting can pull the two apart. That chapter needs a new recording, in a quiet room with nothing playing.
Where can music go in an ACX audiobook?
The submission requirements mention music in one place only: the credits. Credits are the short opening and closing announcements of the title, author and narrator, and ACX asks for them as separate files from the chapters.[3] The rule is one sentence: "You can add music to the credits, but make sure it doesn't obscure the narration."[3] The checklist's credits section covers what each announcement must say.
The credits are files like any other, so the noise floor rule applies to them as well.[3] ACX does not say how it measures a file with music before and after the announcement, so measure the credits files before you upload. ACX's music article asks you to hire a professional editor, or make sure you can "mix and fade" the music and the narration so they complement each other.[1]
Music inside the chapters is allowed too, used sparingly, as long as every file still meets the submission requirements.[1] A bed you can hear under the narration cannot meet them, because every pause carries it.
Common questions
Does ACX allow background music in audiobooks?
Can I keep the music if I turn it down?
Only if the pauses still measure -60 dBFS or lower with the music in them.[3] In a five-minute chapter with a music bed under the whole narration, the pauses measured -35.7 dBFS, 24.3 dB above that limit.
Will mastering fix a noise floor fail caused by music?
No. Mastering cannot separate two sounds that overlap, and in a mixed chapter the music and the voice are one sound. chapterpass lowers the noise in the pauses between phrases and never touches the sound under the voice. Mute the music, export the voice alone as WAV, then master that file.
Does ACX's Audio Lab show the noise floor?
No. ACX says its Audio Lab "does not detect issues such as noise floor or editing errors".[4] Measure the noise floor of each file yourself before you upload it to ACX.
Ready to submit?
Run the free ACX check to see exactly where your file stands, or let chapterpass master it to spec in one step.