ACX Audio Requirements: RMS, Peak, Noise Floor (2026)

ACX is Amazon's route to Audible, and it has the strictest audio specs in the business. Three numbers cause most of the rejections: RMS loudness between -23 and -18 dBFS, true peak below -3 dBFS, and noise floor below -60 dBFS.[1] Get those right (along with format, sample rate, channel, and silence rules) and your files pass everywhere, including Voices by INaudio, Authors Republic, and Google Play.

ACX Audio Submission Requirements vs Other Platforms

ACX is the strictest of the four major platforms. The other three accept wider formats or looser silence windows, but all of them accept files mastered to ACX spec, which is why the ACX values are the ones worth memorising.

ACX[1] requires MP3 at 192 kbps CBR, 44,100 Hz, mono, RMS -23 to -18 dBFS, true peak below -3 dBFS, noise floor below -60 dBFS, 0.5 to 1 s head silence, 1 to 5 s tail silence.

INaudio[4] (formerly Findaway Voices) matches ACX on loudness, peak, and noise floor. Accepts MP3 or FLAC, mono or stereo. Enforces a maximum chapter length. Silence requirements match ACX.

Authors Republic[5] matches ACX on loudness, peak, and noise floor. Accepts MP3, mono or stereo. Requires 1 to 5 s of silence at both head and tail.

Google Play[6] accepts MP3, M4A, FLAC, and WAV at 128 kbps or higher, mono or stereo. Loudness, peak, and noise floor requirements are not published, but files mastered to ACX spec pass without issue.

Master to ACX and your files meet or exceed every other platform's requirements. If you distribute through multiple platforms (and most narrators do), mastering once to ACX saves you from maintaining separate versions for each distributor.

The 8 Specs ACX Measures

ACX runs every chapter file through an automated analysis against these 8 specs.[1] The other platforms use a subset of the same list, so master to these once and your files are ready everywhere.

RMS loudness: -23 to -18 dBFS
True peak: below -3 dBFS
Noise floor: below -60 dBFS
Sample rate: 44,100 Hz
Channels: mono
Format and bitrate: MP3, 192 kbps CBR
Head silence: 0.5 to 1 second
Tail silence: 1 to 5 seconds

These numbers exist because Audible encodes and re-encodes files into multiple formats for different devices.[1] The headroom in peaks, the CBR encoding, and the silence padding all protect your audio from degradation during that process.

What Each Spec Means

RMS Loudness (-23 to -18 dBFS)

The average volume of your narration across the file. Too quiet and listeners strain to hear. Too loud and it distorts or fatigues. Most mastering tools target -20 dBFS. That sits comfortably inside the range: 2 dB below the ceiling, 3 dB above the floor, with room for natural variation between chapters.

If your meter shows a different scale (integrated loudness, VU, or something else), switch it to RMS before comparing against ACX's range.

True Peak (Below -3 dBFS)

The loudest single moment in the file, including peaks that fall between digital samples. The -3 dBFS ceiling leaves room for the signal to expand when re-encoded to lossy formats.[2] Without that headroom, your mastered file can distort on a listener's device even though it measured clean on yours.

Noise Floor (Below -60 dBFS)

The volume of background sound during silence. HVAC hum, computer fans, refrigerators in the next room, and room ambience all contribute. ACX measures the noise floor in the silent sections of your file. If your studio is not acoustically treated, this is usually the first spec to flag. The audiobook noise floor guide covers exactly how to fix it.

Sample Rate (44,100 Hz)

44,100 Hz, the CD standard. Files recorded at 48,000 Hz (common for video work) need to be converted before submission. If your interface supports it, record at 44,100 Hz from the start and skip the conversion entirely.

Channels (Mono)

Mono. Stereo files are rejected by ACX. Audiobooks are spoken word, not spatial audio, so mono is the standard. If you recorded in stereo by accident, collapsing to mono in your DAW is a one-click fix.

Format and Bitrate (MP3, 192 kbps CBR)

Constant bit rate, not variable. CBR ensures the file plays reliably across all devices and apps. Variable bit rate MP3s can cause playback glitches on older hardware and streaming apps, which is why ACX enforces CBR.

Head Silence (0.5 to 1 second)

Room tone (not digital silence) at the start of each chapter. Gives the listener an audio cue that the chapter has begun and lets automated splitting tools locate chapter boundaries cleanly.

Record a few seconds of silence in your booth during the same session and use that as your room tone source. Digital silence (absolute zero) sounds unnatural and can cause audible clicks at the transition to speech.

Tail Silence (1 to 5 seconds)

Room tone at the end of each chapter. Provides a clear pause before the next chapter starts. Same source as your head silence: recorded room tone from your booth, not generated silence.

ACX Noise Floor: Why -60 dBFS?

The -60 dBFS threshold is not arbitrary. It is the level below which background sound becomes inaudible over spoken narration at normal playback volume. At -50 dBFS, a hiss or hum is clearly present between sentences. At -60 dBFS, most listeners cannot hear it even in quiet passages. The 40 dB gap between the speech level (around -20 dBFS RMS) and the noise ceiling preserves the perception of silence in silence.

The practical problem is the gain budget. Raising loudness to reach -20 dBFS raises the noise floor by the same amount. A raw recording at -65 dBFS plus 8 dB of gain lands at -57 dBFS, which fails. That tension, and the approach that solves it, is the subject of the audiobook noise floor guide.

ACX

ACX (Audiobook Creation Exchange) is Amazon's distribution route to Audible, Amazon, and Apple Books. It has the strictest technical requirements of any major platform and the only fully automated spec check. Files are measured against the 8 specs above before any human listens. Every measurement needs to pass. For a full walkthrough of every rejection type, see fix ACX rejection.

Voices by INaudio (Formerly Findaway Voices)

Findaway Voices rebranded as Voices by INaudio in August 2025.[4] INaudio accepts FLAC alongside MP3, accepts stereo files (using Joint Stereo encoding), and enforces a maximum chapter length (ACX does not publish a duration cap). Master to ACX spec and your files pass INaudio without any adjustment.

Authors Republic

Authors Republic accepts both mono and stereo files with a 170 MB per chapter limit.[5] The main difference from ACX is silence padding: Authors Republic requires 1 to 5 seconds at both the head and tail, while ACX allows as little as 0.5 seconds at the head. Authors Republic also requires separate opening and closing credits tracks with the book title, author, and narrator name. You will need to create the credits tracks separately.

Google Play Books

Google Play Books accepts MP3, M4A (AAC), FLAC, and WAV at 128 kbps or higher.[6] Google does not publish specific loudness, peak, or noise floor targets. If your file meets ACX's specifications, it exceeds everything Google Play requires.

Common Questions

Are the ACX audio requirements (RMS, peak, noise floor) the same for audiobook samples?

Yes. ACX applies the same automated spec check to retail audio and to the 1- to 5-minute audition sample. If your sample fails, it fails for the same reason a chapter file would: loudness outside the -23 to -18 dBFS RMS window, true peak above -3 dBFS, or noise floor above -60 dBFS.

What is the difference between sample peak and true peak for ACX?

Sample peak is the loudest single digital sample in the file. True peak estimates the actual analogue signal level between samples using oversampling (ITU-R BS.1770 defines 4x oversampling).[2] The gap is typically 0.5 to 1 dB. ACX measures true peak, so a file reading -3.2 dB on a sample peak meter can still fail.

Can I submit a WAV or FLAC file to ACX?

No. ACX requires MP3 at 192 kbps CBR. WAV and FLAC are accepted by INaudio, Authors Republic, or Google Play, but ACX itself rejects non-MP3 files at automated check.[1]

Why does ACX require mono when listeners have stereo devices?

Audiobooks are spoken word, not spatial audio. A stereo file of a single narrator is two identical channels, doubling the file size for no listening benefit. ACX requires mono to keep catalogue storage and streaming efficient.[1]

What happens if my chapter is longer than the silence window allows?

Silence rules are per file, not per chapter length. ACX requires 0.5 to 1 second of room tone at the head and 1 to 5 seconds at the tail regardless of runtime. If your tail silence is 8 seconds, trim it back. If it is 0 seconds, add room tone recorded from your booth.

Your first file is free at ChapterPass. For quick answers on pricing and privacy, see the FAQ.