ACX Audio Requirements, Explained in Plain English

If you're narrating your own audiobook, or producing one for an author, sooner or later you run into the ACX audio requirements. It's a list of numbers (RMS, peaks, noise floor, bit rate), and if you don't spend your days in a studio, it can read like a foreign language. The good news is that none of it is mysterious. Every requirement exists to make sure listeners get a book that sounds clean, steady, and comfortable to listen to for hours.

I produce audiobooks for publishers alongside my music work, and the same thinking applies whether a book is going through a publisher or you're uploading it yourself. Here's what ACX asks for, what each piece actually means, and how to hit it without guessing.

One note before we start: ACX can update its specs, so check the current requirements on ACX before you upload. The numbers below are the ones ACX publishes as I write this.

The file format: MP3, 192 kbps, constant bit rate, 44.1 kHz

ACX wants each file delivered as an MP3 at 192 kbps or higher, encoded at a constant bit rate (CBR), not a variable bit rate (VBR), with a sample rate of 44.1 kHz.

In plain English: record and edit in full-quality WAV files, then export the MP3s as the very last step. Don't trust your software's default export settings. Open the export dialog and set the bit rate, the sample rate, and constant bit rate yourself. VBR is the one that catches people, because some encoders use it unless you tell them otherwise.

Loudness: RMS between -23 dB and -18 dB

RMS is a measurement of average level. Roughly, it's how loud the file sounds over time, rather than its loudest instant. ACX requires each file to measure between -23 dB and -18 dB RMS.

Too quiet, and listeners turn their volume way up and hear every bit of hiss. Too loud, and it gets tiring. Landing in that window on every chapter is also what keeps the book consistent, so nobody has to reach for the volume knob between Chapter 4 and Chapter 5.

Peaks: no higher than -3 dB

Peaks are the loudest instants: a hard consonant, a laugh, a raised voice in a line of dialogue. ACX says peaks can be no higher than -3 dB.

The catch is that loudness and peaks have to be managed together. If you simply turn a file up until it hits the RMS target, the peaks go over. That's where careful mastering comes in: gentle compression and limiting that keeps the peaks in check so the average level can sit where it needs to, without squashing the life out of the narration.

Noise floor: no higher than -60 dB RMS

The noise floor is what's left when nobody is talking: room hum, computer fans, traffic, the refrigerator in the next room. ACX requires it to be no higher than -60 dB RMS.

This is the requirement that gives home recordings the most trouble, and it's far easier to solve while recording than to fix afterward. A quiet room, good mic placement, and switching off anything that hums will do more than any plug-in. Noise reduction can help, but push it too far and the voice starts to sound watery and processed. Many of the habits from my post on how to record vocals at home carry straight over to narration.

Room tone at the head and tail

Each file needs a little clean "silence" at both ends, which is really room tone (never more than 5 seconds at either end):

• Beginning: 1 to 5 seconds of room tone

• End: 1 to 5 seconds of room tone

That room tone should be free of extraneous sounds. No breaths, clicks, chair creaks, or page turns. It gives each chapter a natural start and finish instead of the voice jumping in or getting cut off.

How the files are organized

ACX is specific about structure, too:

• One chapter or section per file, with the section header read aloud.

• No file longer than 120 minutes.

• Opening and closing credits are included.

• A retail audio sample of 5 minutes or less.

• Consistency across the whole book: the same sound and formatting throughout, and all mono or all stereo files. Don't mix the two.

I also suggest naming each file with its order number plus the section name, so everything stays in the right order when you upload.

My own recording tips

These aren't ACX requirements, just how I'd approach any spoken-word session:

• Record so your voice peaks around -12 dB to -8 dB. That leaves headroom for the louder moments.

• Set the preamp and leave it. If you need more level in your headphones, turn up the monitor level, not the preamp. Changing gain partway through a book is a recipe for chapters that don't match.

• Keep 30 to 60 seconds of clean room tone so you have something to use when editing.

• Edit out the distractions: mouth noise, clicks and pops before or after the performance, and long silent gaps.

How I approach it

My general workflow for a spoken-word project is simple, even if the details take care. Record as clean as possible. Edit carefully for performance and noise. Run quality control on every chapter. Then master to spec. Before anything gets uploaded, I check every file's levels, peaks, and noise floor, and I listen through for the things a meter won't catch.

Meters tell you whether a file passes. Your ears tell you whether it sounds good. The goal isn't only to clear a technical check. It's a book that sounds natural and that a listener can stay with, chapter after chapter, without getting tired.

Need a hand with your audiobook?

If you're working on an audiobook and want help recording, editing, or mastering it to the ACX audio requirements, take a look at my audiobooks and voice-over page. Whether you're starting from scratch or you have chapters that keep failing the numbers, I'm happy to talk it through.

— Jim

If you'd like one-on-one help with your own mix, vocals or audiobook, you can book an online session with me.