Audiobook Loudness and Audio Levels: What Retailers Expect

When you finish an audiobook and go to publish it, the retailer or distributor almost always checks the audio against a set of technical targets before it goes live. These targets cover how loud the file is, how high its peaks go, and how quiet the background is between words. If the file misses them, the upload gets rejected and you are back at your desk figuring out what "RMS out of range" means. This post walks through the numbers most platforms expect, how a file exported from AudioProducer.ai lines up with them, and what to do if a retailer sends a file back.

Why loudness specs exist for audiobooks

Audiobooks are listened to in cars, on walks, over kitchen speakers, and through cheap earbuds. A listener who sets a comfortable volume during chapter one does not want to lunge for the volume knob in chapter two. Loudness specs exist so that every file in a catalog plays back at a consistent level and stays clear of the distortion that happens when a signal is pushed too hard.

Retailers also care about the quiet parts. A noisy background, hum, or hiss is far more obvious over a long listen than it is in a thirty second sample, so platforms set a ceiling on how loud the silence between words is allowed to be. Meeting these targets is not about making the file louder for its own sake. It is about landing inside a predictable window so the file passes review and sounds even next to everything else the listener owns.

The common RMS, peak, and noise-floor targets

Three measurements come up again and again. Knowing what each one describes makes the retailer requirements much easier to read.

• RMS (average loudness). This is the overall perceived loudness of the file, averaged over time. The widely cited ACX range is an RMS between -23 dB and -18 dB. A file that sits in this band feels neither faint nor overbearing.
• Peak level. This is the single loudest instant in the file. ACX asks that peaks stay at or below -3 dB. Keeping headroom below 0 dB prevents the hard clipping that turns a loud consonant into a click or crackle.
• Noise floor. This is how loud the quiet passages are. The common ceiling is -60 dB RMS or lower, meaning the background between words should be very close to true silence.

Different platforms word these slightly differently, and some measure loudness in LUFS rather than RMS, but the underlying idea is the same: a target average, a peak ceiling, and a clean noise floor. Always read the spec for the exact store you are publishing to, because the numbers are the contract.

How exported files line up with those specs

AudioProducer.ai generates your narration and gives you a finished MP3 to download. Because the audio is synthesized from a clean text-to-speech pipeline rather than recorded through a microphone in a room, two of the three problem areas are handled before you ever open the file. There is no room tone, no air conditioner, and no desk bump, so the noise floor on an exported file is effectively silent and clears the -60 dB expectation comfortably. The generated speech is also produced at a controlled level, so peaks are not slamming into 0 dB the way a live recording sometimes does.

What we want to be precise about: AudioProducer.ai exports the audio file. It does not distribute, host, or publish that file to Audible, Spotify, Apple, ACX, libraries, or any podcast feed, and it does not submit your book to a retailer on your behalf. You download the MP3 and publish it wherever you already publish. The retailer specs above are informational so you know what the file you export should meet, not a guarantee that any given store will approve it.

Checking levels before you upload

You do not have to trust that a file is in range. You can measure it in a couple of minutes with free tools before you submit anything.

• Open the file in an audio editor. A free editor such as Audacity can analyze a whole file and report its RMS, its peak amplitude, and, if you select a silent gap, the level of the noise floor.
• Read the three numbers. Confirm the average sits inside the retailer's RMS or LUFS window, the peak is under the ceiling (for example -3 dB), and a silent stretch measures below -60 dB.
• Apply gentle correction only if needed. If the average is a touch low, a small normalization or gain move brings it into range. Nudge the level to the target rather than pushing hard for maximum loudness, which is what causes rejections in the other direction.

Doing this once for a chapter tells you a lot about the whole export, since a synthesized file is consistent from start to finish. Measure, confirm, and move on.

What to do if a retailer rejects a file

Rejections almost always name the measurement that failed, and each one has a plain fix.

• "RMS too low" or "too quiet." Apply a modest gain increase or normalize to the middle of the required range, then re-measure.
• "Peaks exceed the limit." Lower the overall level slightly, or apply a peak limiter set to the ceiling, so the loudest moment drops under -3 dB without squashing the rest.
• "Noise floor too high." This is rare with a synthesized export, but if it appears, check that you did not add hiss during your own editing or a lossy re-encode, and re-export cleanly.
• Format or bitrate issue. Sometimes the loudness is fine but the container is not. Confirm you sent the format the store asked for. Our post on MP3 versus M4B audiobook formats covers which to choose.

Make the specific correction the retailer named, re-run your three measurements, and resubmit. Because you control the file, you can adjust and re-upload as many times as it takes without starting the narration over.

Frequently asked questions

New to audiobook audio levels? These are the questions we hear most often.

If you are still setting up your first project, start with our guide on how to make an audiobook with AI, then see how to publish an AI-narrated audiobook and how to publish on Findaway Voices once your file passes its level check. Once levels are set, audiobook metadata and ID3 tags and whether you need an ISBN for an audiobook are the next things to sort before you upload.