Mastering an ACX Fail

So end of last March (before the RMS Normalize plugin came out), I finally got my first chapter recorded. I did my edits and followed Mr. Kozikowski’s most excellent instructions, using the SetRMS plugin, then the Limiter. It passed ACX Check and sounded good, and I sent it off. Aside from the issues involved in getting to the sample submission page and then getting them to download it before the link vanished, then got around to letting me know that I needed to re-submit, I eventually got their comments back. I did make one editing error that will be insanely easy to fix. I just have to do it. However, I got the below comment and it has me perplexed.

“My only other comment is that this file is a bit too dynamic. Consistency in audio levels, tone, noise level, spacing, and pronunciation gives the listener an enjoyable experience. Drastic changes can be jarring to the listener and distract from the listening experience. Review the tutorials at the links provided below.”

The links in question led to these two videos which were only moderately helpful: and

The second one is a recording of a class and there is no link to the manual referred to.

So both videos talk about Equalizing, Normalizing and Compressing, which, if I’m understanding Mr. Kozikowski’s advice, appear to equate to Equalize LF RollOff (which I did not do in my sample because it passed ACX Check), SetRMF (which would now be RMF Normalizer), and Limiter. I’ve uploaded about 30 seconds (apologies for the longer length, but I’m hoping you’ll hear some of the supposed dynamics issue in it) from the sample. Should I have run the Equalizer? it sounds like from the note that it’s a mastering issue rather than a performance one. Or should I try running the raw clip through Audacity’s equalizer, normalizer and compressor effects?

Trying not to tear my hair out.

I’ve been working on a new tool which aims to give more information and alert the producer to ‘possible issues’ rather than a cozy pass/fail ‘answer’.

Note that the labels do not necessarily mean that the audio is “bad” or “wrong”. They just indicate that the producer should review those sections.

This is what it shows:

It’s been fun keeping up with this. As I posted on a sister message, I wrote an Audiobook Mastering paper which is now about 2/3 out of date.

SetRMS would set loudness in many different versions of Audacity, but was “oil-based paint drying slow.”

RMS Normalize is much better behaved and much faster…on Audacity 2.1.3. They do the same job (set RMS Normalize to -20, not -18). They set the overall RMS (volume).

I posted a disclaimer. All this work and these tools are to get you past Technical Conformance. That’s what kills most people. I’m too loud. I’m too noisy, I have clipping, etc.

ACX has a second layer that deals with Theatrical Conformance. That’s where you find out your voice is so bad it qualifies as a non-lethal weapon by the State of California. And that’s also where you find it’s a bad idea to wander in volume, expression, and sibilance through the course of the show. Theatrically Expressive is not welcome.

Sometimes we can come up with patches, adjustments or even software packages to help, but if you find yourself correcting a book word by word (I’m not making that up) it’s time to read it again.

Yes, that’s a standing joke. One poster made it through all the quality conformance tests and was bounced for including the wrong length of silence in a clip.



Equalizing, Normalizing and Compressing,

Ahhh. No. They’re talking about changing the quality of your voice. If I did it right, all the “Mastering” tools leave you sounding the same as you started out. The LF-Rolloff thing appears in Equalization, but primarily affects thunder, trucks passing by and earthquakes. It may affect your voice if you have one of those old-time radio, rumbly voices, but otherwise, no.

Normalizing, at least the way we do it is a volume control. Full stop. It’s a sister to Amplify. You tell them what you want in different ways, but they’re both just one pass volume controls.

Their Equalizing is to crank up the bass or make your voice more sparkling or crisp. Compression is the one you probably really need. That’s the one that changes volume down the timeline depending on how loud you are. It can make your voice more dense or forward.

Please note (underline three times) ACX has a failure called “Overprocessing.” These vocal processing tools are hard to use. If ACX can hear what you did, you’re dead.


What? We’re supposed to be reading in a monotone? That sounds like it would be hellishly boring to listen to.

Not what you said they said. It’s a compromise between being interestingly dynamic, and jarringly dynamic.

“My only other comment is that this file is a bit too dynamic. Consistency in audio levels, tone, noise level, spacing, and pronunciation gives the listener an enjoyable experience. Drastic changes can be jarring to the listener and distract from the listening experience.”

We’re supposed to be reading in a monotone?

No. That’s my trick. That’s why nobody is beating down my door to read for them.

Reading with pitch changes is one of the things you can do. Just don’t bellow or whisper without taking precautions not to have the recorded volume change too much.

You may recognize limited volume change presentation. Broadcast Radio does that. They have equipment at the antennas on the hill to squeeze most of the volume changes out of a show. They are legally required to keep presentation volume within limits. That allows you to listen in the car without constantly changing the volume on the radio. It lets me hike around the neighborhood without constantly fiddling with my Personal Music Device.