Common ACX Rejection Reasons and How to Fix Them

Getting your audiobook rejected by ACX is frustrating. You've invested hours in recording and editing, submitted what you thought was a finished product, and now you're staring at a rejection notice with technical jargon you may not fully understand.

Here are the most common reasons audiobooks fail ACX quality control, ranked roughly by how often they cause rejections, along with concrete fixes for each one.

1. RMS Level Out of Range

The rejection: Your file's average loudness falls outside the -23 to -18 dBFS range. This is the single most common ACX rejection.

Why it happens: Most home studio recordings come in too quiet, typically around -30 to -40 dBFS. Narrators who are careful not to clip (a good instinct) often record too conservatively. Less commonly, files are too loud because of aggressive normalization without proper limiting.

How to fix it: Measure your file's RMS level. If it's below -23 dBFS, apply gain to bring it up. If the dynamic range is too wide (causing some sections to be way too loud when the average is in range), apply gentle compression (2:1 ratio, slow attack) before adjusting gain. Target -20 dBFS for comfortable margin.

After loudness adjustment, you'll almost certainly need to re-check your peak levels—bringing up the average pushes peaks higher too.

2. Peaks Above -3 dBFS

The rejection: One or more inter-sample peaks exceed -3 dBFS.

Why it happens: After loudness normalization, transients (plosives, mouth clicks, emphasis on certain words) get pushed above the ceiling. Standard sample-peak limiters can miss inter-sample overs, which is why ACX uses true peak measurement.

How to fix it: Apply a true-peak-aware brickwall limiter with a ceiling of -3 dBFS (or slightly below, like -3.1 dBFS, for safety margin). The limiter should operate on the decoded analog signal, not just sample values. Run it after loudness adjustment but before MP3 encoding.

If your limiter is reducing gain by more than 2-3 dB on most peaks, back off your loudness normalization. Heavy limiting damages the natural sound of narration.

3. Noise Floor Too High

The rejection: Background noise during silence exceeds -60 dBFS.

Why it happens: Room noise from HVAC, computers, traffic, and electrical hum adds up. What sounds silent to your ears in a quiet room might measure at -50 dBFS. The problem gets worse after loudness normalization because boosting the signal also boosts the noise.

How to fix it: A noise gate cleans up silence between phrases by dropping the level when you're not speaking. For persistent noise that's audible during speech, spectral noise reduction can help, but use it sparingly—heavy noise reduction creates a characteristic "underwater" artifact that's worse than mild background noise.

The best long-term fix is improving your recording environment. A quiet room with acoustic treatment produces recordings that need minimal noise reduction.

4. Wrong Format or Bitrate

The rejection: File is not MP3, not 192 kbps, not CBR, wrong sample rate, or not mono.

Why it happens: Encoding settings get overlooked. Common mistakes include exporting as WAV (forgetting the MP3 step), using VBR instead of CBR, leaving the sample rate at 48 kHz, or forgetting to downmix stereo to mono.

How to fix it: This is the most straightforward rejection to fix. Export explicitly as MP3, 192 kbps CBR, 44.1 kHz, mono. Verify the output file's properties with a tool like MediaInfo or ffprobe rather than trusting your DAW's export dialog. Some DAWs have bugs in their export settings that don't actually apply what the UI shows.

5. Missing or Wrong Silence

The rejection: Head silence is too short, too long, or missing. Tail silence is outside the 1-5 second range.

Why it happens: Narrators often start speaking immediately at the top of the file with no lead-in, or they trim silence too aggressively during editing. Some editors cut silence to zero-length padding, which creates no audible gap.

How to fix it: Add 0.5 to 1 second of room tone at the beginning and 1 to 5 seconds at the end. Don't use digital silence (absolute zero) because the transition from nothing to room tone is audible and unnatural. Use actual room tone from your recording session, or generate low-level pink noise that matches your ambient level.

Crossfade between the silence padding and the speech content. Even a short 50ms crossfade prevents clicks at the transition point.

6. Inconsistency Across Chapters

The rejection: Chapters have significantly different loudness levels, noise floors, or tonal characteristics.

Why it happens: Recording sessions spread across multiple days or weeks with different mic positions, room conditions, or equipment settings. Processing chapters individually with different settings rather than batch processing with identical parameters.

How to fix it: Process all chapters through an identical mastering chain. Use the same compression settings, the same limiter ceiling, and the same noise treatment for every file. Batch processing tools are essential for this—manual per-chapter adjustment inevitably introduces variation.

If chapters were recorded under very different conditions (new microphone, different room), you may need to EQ-match them first to get a consistent tonal baseline before applying the rest of the chain.

7. MP3 Encoding Shifts Levels

The rejection: Your WAV master passes all checks, but the final MP3 doesn't.

Why it happens: MP3 encoding is lossy, and the encoding process can raise peak levels by 0.5-1 dB. A file with true peaks at exactly -3.0 dBFS before encoding might hit -2.5 dBFS after encoding. This is a subtle issue that catches many producers who verify their WAV files but not their MP3 output.

How to fix it: Always verify the final MP3, not just the source. Set your limiter ceiling slightly below -3 dBFS (e.g., -3.1 dBFS) to leave margin for encoding-induced level changes. Better yet, decode the MP3 back to PCM and measure its true peaks to confirm compliance.

Avoiding Rejections Entirely

The common thread across all these rejections is that each one requires specific measurement, specific tools, and specific verification—multiplied by every chapter in your audiobook. Missing any single checkpoint on any single chapter means rejection.

ACX Pass was built to eliminate this entire category of problems. It runs all 8 compliance checks on every file and only exports files that pass every check. If something fails, you get a specific report telling you which check failed and what the measured value was, so you know exactly what to fix in your recording.