A caution. ACX has a failure called “Overprocessing.” That’s why the audiobook mastering processes were all chosen so you can’t hear them running. If you need a compressor to even out your work, it’s possible your recordings are too “wild” or uncontrolled and fixing that may do a world of good.
If you’re in Windows, it’s pretty common for other software packages to “help you” in the background, and you end up with sound effects fighting each other. That can require a lot of coffee to fix.
Also, anything you do to the first chapter, you need to do to all the chapters in that book. ACX requires your chapters to match. There is no applying effects or corrections only “when needed.”
You’re still invited to post a sample. We can’t give you a neat list of all the things you can do wrong, some more obvious than others.
A limiter is a kind of fast dynamic compression. The limiter will “push down” the peaks with very little effect on the RMS level and no effect on the noise level.(1) The Audacity compressor is very good!
(1) That assumes no make-up gain. You don’t want make-up gain because you that will re-boost your peaks (and boost everything else) and you want your peaks at -3dB or less (more negative), and presumably you’ve already adjusted the RMS level.
Do you have special circumstances that require a compressor for spoken word performances?
You are cautioned about applying peak limiting such that maximum peak loudness is exactly -3dB. Yes, that’s the ACX specification, but the conversion from an Audacity Production to an MP3 Submission while good, is not exact. It is absolutely possible to produce an MP3 with a peak loudness of -2.8dB which does not pass. The mastering system uses a goal of -3.5dB which makes errors like that much less likely.
I hope this response reaches all of you who have provided advice: Thank you.
I am an actor, not an engineer, and am still getting my head around the concept of negative numbers meaning greater volume.
It turns out my problem was caused by double-processing the track. I’ll spare you the details, but I was applying an equalization, RMS normalization, limited MACRO I copied from one of the first tutorials I encountered, then reapplying most of them. As I said … not an engineer.
I tried to post my most recent (and successful) effort for your observations, but new members are not allowed to post.