ACX Loudness Requirements: Pass Audible's Audio Spec Every Time

ACX rejects audiobook submissions for the same four reasons over and over. RMS too quiet. Peaks too hot. Noise floor too loud. No room tone at the head and tail. The narrator gets the rejection email three to seven days after submitting, has to remix every chapter, and resubmits hoping it passes the second time.

This is one of the most expensive workflow problems in audiobook narration. A 12-hour book that gets rejected costs you 3 to 4 hours of extra mixing work. Multiply that across a career and you're losing weeks per year to bounces that didn't need to happen.

This guide covers the four ACX audio specs in detail, the common mistakes that get audio rejected, and a pre-submission checklist that takes 90 seconds and catches every spec violation before you upload.

The Four ACX Audio Specs

ACX publishes its audio submission requirements publicly. There are four numbers every chapter has to hit. If any one of these is off, the chapter fails QC and the audiobook goes into 'corrections needed' status.

1. RMS between -23 dB and -18 dB

RMS (Root Mean Square) is the average loudness of your audio over time. ACX wants this between -23 dB and -18 dB. The sweet spot working narrators target is around -20 dB.

Too quiet (below -23 dB) and the listener has to crank the volume to hear you, which means every breath and background sound becomes audible. Too loud (above -18 dB) and the audio sounds compressed and harsh.

2. Peak no higher than -3 dB

Peaks are the loudest moments in your audio. ACX wants the absolute peak to be no higher than -3 dB across the entire chapter. This leaves headroom for the platform's processing.

If you have a single 'p' pop or a hard consonant that spikes past -3 dB, the entire chapter fails. The fix is either a limiter set to -3 dB or careful pop filtering during recording.

3. Noise floor at -60 dB or lower

Noise floor is the background hum, hiss, or ambient sound that's present in your recording when no one is speaking. ACX requires this to be at -60 dB or lower (further from 0). Anything louder fails.

This is where most home studio narrators get tripped up. Your AC unit, your computer fan, traffic from outside, even your dog's breathing two rooms over can push noise floor past the limit. The fix is acoustic treatment and a clean signal chain, not noise reduction plugins (those introduce artifacts that ACX also dislikes).

4. 0.5 to 1.0 seconds of room tone at the head and tail

Every chapter needs between half a second and one full second of clean room tone (just the silence of your booth, no voice) at the very beginning AND end of the file. This lets ACX stitch chapters together cleanly during playback.

Skip this and the chapter gets rejected even if every other spec is perfect. Add too much (more than 1 second) and it also fails. Set a hard rule of exactly 0.7 seconds on both ends and you're always within spec.

The Most Common Reasons ACX Rejects Audio

After reviewing dozens of ACX rejection notices, the same five issues come up over and over.

1. Noise floor too high. The number one reason. Your home studio isn't quiet enough, or you're using noise reduction that's introducing background hiss.
2. Inconsistent RMS across chapters. Chapter 1 averages -20 dB. Chapter 7 averages -16 dB because you recorded closer to the mic that day. ACX wants consistency across the whole book.
3. Mouth noise and clicks not removed. Saliva sounds, lip smacks, and tongue clicks at -55 dB count as part of the noise floor calculation. Edit them out.
4. Plosives clipping past -3 dB. A single 'p' pop in chapter 4 will fail the whole chapter. Use a pop filter and/or a limiter set to -3 dB.
5. Missing or wrong-length room tone. Either no room tone at the head/tail, or 2-3 seconds instead of the required 0.5-1.0. Both fail.

How to Hit Every Spec on the First Try

The workflow that working ACX narrators use looks like this.

1. Record clean. Treat your room properly. Aim for a recording with as low a noise floor as possible from the start. Fix problems at the source rather than in post.
2. Edit before normalizing. Remove breaths, mouth clicks, false starts, and excessive silence first. Then check noise floor.
3. Normalize to RMS -20 dB. Most DAWs have a loudness normalize function. Set the target to -20 dB RMS as a safe middle of the ACX range.
4. Limit peaks to -3 dB. Apply a brick-wall limiter set to -3 dB. This catches any hard consonants or plosives that spike too hot.
5. Add room tone. Hard-cut to exactly 0.7 seconds of room tone at the head and tail. Use the room tone you recorded at the start of each session.
6. Run a final spec check. Use a tool that measures all four specs (RMS, peak, noise floor, head/tail silence) and confirms each passes before you upload.

The 90-Second Pre-Submission Checklist

Run this before you upload any chapter to ACX. It catches 99% of rejection-worthy issues.

1. RMS in your DAW shows between -23 dB and -18 dB. Aim for -20 dB.
2. Peak meter never goes above -3 dB. Scan the whole file. Look for spikes.
3. Noise floor (the silent passages) is at -60 dB or lower. Solo a silent section and check the meter.
4. Room tone is present at the very start AND the very end. Length: 0.5 to 1.0 seconds. No more, no less.
5. File is in the right format. 192 kbps or higher CBR MP3, 44.1 kHz sample rate, mono (unless you've negotiated stereo with your producer).
6. Chapter is the right length. ACX wants chapters between 5 seconds and 120 minutes. Most books split chapters into 30 to 60 minute files.

The Specs ACX Doesn't Publish But Cares About

Beyond the four published specs, there are quality factors that affect approval but aren't in the official documentation.

• Consistency across chapters. Even if each chapter individually passes, if your RMS varies by more than 2 dB from chapter to chapter, the QC reviewer will flag it as inconsistent.
• Sibilance and harsh esses. ACX doesn't reject for harsh 'S' sounds explicitly, but if every 'S' is painful to listen to, the QC reviewer will note it as a problem.
• Edit point audibility. Cuts between phrases that aren't smoothed out can cause audible pops. Use crossfades or zero-crossing edits.
• Mouth noise pacing. Even if mouth noise is below -60 dB, if there's a click every other word, it counts as a quality issue.

Tools That Help Hit ACX Spec

You don't need expensive plugins to hit ACX spec. The basics that most working narrators use:

• iZotope RX for noise reduction and click/mouth noise removal. The industry standard.
• Waves WLM Plus for loudness metering. Shows all four ACX specs at once.
• FabFilter Pro-L 2 for brick-wall limiting. Set to -3 dB ceiling.
• Auphonic as an alternative loudness leveler. Subscription-based.
• VoiceEditSuite Loudness Normalizer with the ACX preset. Drop in the file, get back a spec-compliant export.

What to Do If ACX Rejects Your Submission

If you get a rejection notice, ACX usually tells you which spec failed. Read it carefully. Common notices and what they actually mean:

• 'Audio is too soft' = Your RMS is below -23 dB. Re-normalize to -20 dB.
• 'Audio is too loud or distorted' = Your RMS is above -18 dB OR your peaks exceed -3 dB. Pull RMS back, apply a -3 dB limiter.
• 'Excessive noise' = Your noise floor is above -60 dB. Treat your room better, or apply careful noise reduction.
• 'Missing or excessive head/tail silence' = Self-explanatory. Trim to 0.7 seconds head and tail.

Hit Spec, Submit Once, Move On

ACX rejection is preventable. The four specs are public, the workflow is standardized, and the tools exist. Most narrators who get rejected aren't doing anything wrong fundamentally. They're just skipping one step in the pre-submission check.

Build the 90-second checklist into your workflow. Run it on every chapter before you upload. The hours you save per book add up to weeks per year. While you're tightening up your post-production, check out our voice over editing guide and the voice over home studio setup guide for adjacent workflow improvements.