I Built a Tool to Automate the Part of Audiobook Mastering Nobody Talks About
A narrator reached out a few months ago. She was weeks into her first audiobook project — solo narration, self-published, doing everything herself. The performance was strong. She had put real work into pacing, character voices, the emotional shape of each chapter. The kind of effort that makes you want to help someone get across the finish line.
Then ACX rejected her files.
RMS levels too high. Peak ceiling exceeded. Noise floor out of spec. She had no idea what any of that meant. She had spent weeks recording and editing, and now a set of technical requirements she had never encountered was standing between her and publication. She was ready to walk away from the whole project.
I told her I could fix it. And I could — I knew exactly what needed to happen to each file. Normalize the loudness. Limit the peaks. Check the noise floor. Add the right silence padding at the head and tail of every chapter. Export to the exact format ACX expects. None of it was difficult. All of it was tedious.
Her book had thirty-two chapters. I set up a session, processed each one, verified it against the spec, and delivered the files. It took the better part of two weeks between other projects. Not because any single chapter was hard, but because it was the same sequence of steps, thirty-two times. Load the file. Measure. Process. Measure again. Export. Next.
That project got published. She was happy. I moved on.
Then another narrator reached out with the same problem. Different book, different voice, different genre — same technical failures. The audio was good. The compliance wasn't. Same fixes. Same repetitive work.
It kept happening. A narrator with a clean, well-edited recording that just needed to meet a set of numbers. An indie author who hired someone on Fiverr for narration and got files back that sounded fine but failed every ACX check. A podcaster pivoting to audiobooks who didn't realize the technical bar was different.
The pattern was obvious. The audio quality was already there. The performance was already there. What was missing was a mechanical step — adjusting levels, limiting peaks, verifying specs, reformatting — that had nothing to do with creativity or skill. It was compliance work. Repetitive, predictable, and identical every time.
I kept thinking about those mastering sessions and ear fatigue conversations I'd had — how the real cost of repetitive technical work isn't the time, it's the attention it drains from decisions that actually matter. This was the same thing. Hours spent on mechanical processing that a script could do in seconds.
So I built the script. Then I rebuilt it properly. Then I thought about every narrator and author who had contacted me with this exact problem and realized they shouldn't need to contact anyone at all. The whole thing should be a tool someone can use without knowing what RMS stands for.
That became ChapterPass ↗.
What ChapterPass Does
ChapterPass ↗ takes your finished audiobook chapters and makes them meet ACX technical requirements. You drop files in, it processes them, you get compliant files back. Here is exactly what it handles:
- Loudness normalization — ACX requires audio between -23 and -18 dBFS RMS. That is a measure of perceived loudness across the whole chapter, not just the loud parts. ChapterPass adjusts each file to land in that range.
- True peak limiting — Peaks must stay below -3 dBFS. These are the instantaneous loudest moments in the audio. If they are too high, playback devices clip and distort. ChapterPass catches and limits them.
- Noise floor check — Must be below -60 dBFS. This is the level of background noise between sentences — room tone, hiss, hum. ChapterPass measures it and flags files that do not pass.
- Silence padding — ACX requires specific amounts of silence at the beginning and end of each chapter. ChapterPass adds the correct head and tail padding automatically.
- Format conversion — The final file must be mono, 44100 Hz sample rate, MP3 at 192 kbps CBR. ChapterPass converts to this exact specification.
What It Doesn't Do
This matters more than the feature list.
ChapterPass will not fix a bad recording. If your audio has room noise, electrical hum, echo, mouth clicks, or distortion baked into it, those problems exist before compliance and they will exist after. No amount of loudness normalization fixes a noisy room.
It will not guarantee ACX acceptance. ACX uses human reviewers who evaluate audio quality, consistency, and production standards beyond the raw technical specs. Meeting the numbers is necessary but not sufficient.
It will not replace the work of performing, recording, and editing well. Those are skills. They take time to develop and they are what make an audiobook worth listening to.
ChapterPass handles compliance, not rescue. If your audio is already well-recorded and cleanly edited, it gets you the rest of the way. If your audio has fundamental problems, you need to fix those first — or hire someone who can.
Your Audio Stays on Your Machine
Everything runs locally in your browser using WebAssembly. Your audio files never leave your computer. Nothing gets uploaded to a server. There is no account, no cloud processing, no third party touching your work.
This was a deliberate decision. Narrators and authors are protective of their unreleased work — they should be. I did not want anyone to have to trust a server with their audiobook chapters before publication. The processing happens on your machine, in your browser, and the files stay yours.
I have written before about automation in mastering and how the interesting question is not whether machines can do the work, but which parts of the work they should do. Audiobook compliance is one of those parts. It is purely mechanical. It requires zero creative judgment. It is the same operation every time. That is exactly the kind of work that should be automated, so the human can focus on the work that actually requires being human.
Your first chapter is free. Drop a file in, see what comes back. ↗