Studio credit
Elephant Audio
Audio production · Sound design · Music production
LahorePKEngineering sound since 2018
Back to the reference index
Platform reference
ACX / Audible — Audiobook Submission Requirements
What audio specification does ACX actually require?
ACX requires each audiobook file to sit between −23 and −18 dB RMS, peak below −3 dB, and hold a noise floor below −60 dB RMS. Files are 192 kbps or higher CBR MP3 at 44.1 kHz, all mono or all stereo, with one to five seconds of room tone at head and tail, and no file longer than 120 minutes.
Source
Transcribed from the published standard, read 2026-09-19. Check it yourself
−23 to −18 dB RMS
required volume
a window, not a target
−3 dB
peak maximum
“peak levels are less than −3dB”
−60 dB RMS
noise floor ceiling
the one most submissions fail
1–5 s
room tone
at the head and tail of every file
The specification, as ACX writes it
Read from ACX's own submission requirements, 2026-09-19:
- Volume is between −23dB and −18dB RMS.
- Peak levels are less than −3dB.
- Noise floor is less than −60dB RMS.
- 192 kbps or higher CBR, 44.1 kHz MP3.
- All files must be in the same channel format — all mono or all stereo.
- Between 1 and 5 seconds of room tone at the beginning and end of each file, and it must not exceed 5 seconds.
- Each file must be no longer than 120 minutes; audio samples must be 5 minutes or less.
These are not LUFS, and the difference is not pedantry
Almost every other delivery figure in this reference is stated in LUFS or LKFS — a loudness measurement weighted to approximate how human hearing responds, integrated across a whole programme. ACX specifies dB RMS, which is a plain average of signal power with no weighting at all.
So a file can measure correctly in one and badly in the other, and a meter set to the wrong mode will pass work that ACX rejects. ⚠️ The numerals invite the mistake: −23 dB RMS and −23 LUFS look identical written down and describe different things. Check which mode the meter is in before deciding the file is finished.
The noise floor is the spec that actually fails people
The volume window is straightforward — it is a gain decision, and any competent mastering pass lands inside it. Peak is a limiter setting. The noise floor is different in kind, because it does not measure the recording. It measures the room the recording was made in.
Below −60 dB RMS means the silence between sentences has to be genuinely quiet: no computer fan, no fridge compressor cycling two rooms away, no traffic through single glazing, no air conditioning. None of that is fixable by processing without taking the voice down with it — broadband noise reduction aggressive enough to reach −60 leaves a performance sounding thin and slightly underwater, which is its own rejection reason.
That is why this is the specification that sends work back. It is a property of the space and the microphone placement, decided before anyone pressed record.
⚠️ Who can hold an ACX account, and who cannot
ACX is open to rights holders and producers resident in the United States, the United Kingdom, Canada and Ireland, with a local taxpayer identification number and banking details in one of those four countries.
That is a constraint on the ACCOUNT, not on the work. An engineer preparing files that meet this specification can be anywhere; the person submitting them cannot. Anyone planning to publish on ACX from outside those four markets should establish that before budgeting for the audio.
Sources
- ACX — Audio Submission Requirements— Every figure on this page. Read 2026-09-19.
Common questions
Why does ACX reject files that sound fine?
Usually the noise floor. A recording can sound clean on headphones and still carry a room tone above −60 dB RMS, because the ear stops noticing steady background noise within seconds while a meter does not. It is measured in the silence between sentences, so the fix is the room and the mic placement rather than the take.
Is −23 dB RMS the same as −23 LUFS?
No. RMS is an unweighted average of signal power; LUFS is weighted to approximate human hearing and integrated over time. The two can differ substantially on the same file. A meter left in the wrong mode is a common reason work passes in the studio and fails on submission.
Can the whole book be one file?
No — each file must be no longer than 120 minutes, and every file in a title has to share the same channel format, all mono or all stereo. Mixing mono and stereo files across a title is a rejection on its own, independent of how each one measures.
Does an engineer outside the US, UK, Canada or Ireland rule this out?
Not for the audio work. ACX restricts who can hold an account and be paid, not who can prepare the files. The rights holder or producer must be resident in one of those four countries; the person mastering to the specification can be anywhere.
Next
- A platform that publishes guidance rather than a gate
- How loudness is measured, where LUFS applies
- Delivery specs versus playback normalisation
- Mastering to a published delivery specification
This page is the lookup. Broadcast and Streaming Loudness Specs for Commercials is the explanation behind it.
Preparing audio for ACX?
The volume window and the peak ceiling are settings. The noise floor is a room, and it is the one worth measuring before the recording rather than after the rejection.
Last updated: