ACX (Audiobook Creation Exchange)
ACX is Amazon's Audiobook Creation Exchange, the marketplace and production platform that distributes finished audiobooks to Audible, Amazon and Apple Books. It is where most English-language self-published audiobooks are submitted, and its technical specification is the de facto standard the rest of the industry is measured against.
The specification ACX enforces
Every file submitted through ACX is checked against a fixed audio spec before a human ever hears it. A file that misses any one of these is rejected automatically:
| Measure | ACX requirement |
|---|---|
| RMS (average loudness) | Between -23 dBFS and -18 dBFS |
| Peak level | No higher than -3 dBFS |
| Noise floor | No higher than -60 dBFS |
| Sample rate | 44.1 kHz minimum |
| Bitrate | 192 kbps constant, MP3 |
| Per-file length | Under 120 minutes |
| Room tone | 0.5 to 1 second at the head and tail of every file |
Those seven numbers are the whole gate. They are also why the phrase "ACX compliant" is used as shorthand for technically acceptable audio well outside Amazon's own ecosystem.
What ACX is not
ACX is a distribution and rights marketplace, not an audio standard body and not a narrator directory in the ordinary sense. Two distinctions authors regularly conflate:
- ACX is not Audible. Audible is the retail storefront. ACX is the submission and rights layer feeding it.
- ACX is not the only route to Audible. Aggregators and direct distributors also reach the same catalogue, usually without the exclusivity terms ACX offers in exchange for a higher royalty rate.
ACX and AI narration
ACX has historically required human narration, and its terms have shifted more than once as AI narration became common. Treat any blanket claim that ACX does or does not accept AI-narrated audio as time-sensitive: check the current terms rather than a blog post, including this one. What has not changed is the audio specification above, which applies identically no matter how the audio was produced.
What ACX checks beyond the audio
Three non-audio requirements catch more first submissions than the specification does. Every book needs opening credits naming the title, author and narrator, and closing credits at the end. Every book needs a retail sample of one to five minutes, chosen to represent the narration rather than to hook the reader. And every audio file has to correspond to a section of the book, so a single file containing three chapters is rejected even if the audio is flawless.
Why it matters when you are choosing a producer
A producer that cannot hand you files meeting the seven measures above has handed you a distribution problem, not a finished audiobook. TomeVox delivers to that specification and does not require distribution exclusivity, so the same files can go to ACX or anywhere else.