audio book--RMS level at 25.7 --I need it at about 20

Hi there,

I’m having difficulty understanding how to get the RMS levels required by ACX to the right levels. Sometimes all is fine. Other times, not so much.

Right now I have the maximum amplitude set to -1.0db. And I have a wave stat add on that tells me the RMS is 25.7. ACX says it has to be between -18 and -23. No matter what I do I can’t get it down because the amplification and new peak boxes won’t let me put in any values at all. So I’m sitting here staring at the screen just not understanding why I can fix this. I know it’s because of something I haven’t learned yet…

I’m not a professional as you can probably tell…

Thanking you in advance,

Right now I have the maximum amplitude set to -1.0db. And I have a wave stat add on that tells me the RMS is 25.7. ACX says it has to be between -18 and -23.

I’ve forgotten the ACX requirements. Is -1dB peak OK?

Go to Amplify and click the box to Allow Clipping. Then Amplify by +3dB. Now you’ll have -22.7dB RMS and +2dB peak (which is potential clipping, but it’s not actually clipped unless you export).

Nex run the Limiter set to Hard Limit at -1dB. (That should get you in the range, although the RMS will drop a little when you limit.)

“measure between -23dB and -18dB RMS and have -3dB peak values and a maximum -60dB noise floor”

You can try Audiobook Mastering version 4.

It boils down to three Audacity effects tools and a custom measuring tool.

If you’re work is relatively quiet, it’s one pass through the three tools and out the door. Please don’t ad-lib or scramble the tools, they work in order. If you’re already close, they will give you an ACX reading in the middle of the specifications, match and still sound like you.


The whole last half of that document is what happens when you fail noise, a very common problem for home production.

ACX has automated testing for the specifications we’re adjusting, but the second half is their Human Quality Control. You have to sound natural with reasonable background noises. No cellphone honking or other artifacts. Your chapters need to match each other.

They have a failure called “overprocessing.”

They also have specifications for silent segments before and after, chapter announcements, etc. etc.

One presenter got bounced when she made it perfectly through both quality control passes and then got bounced for the wrong number of silent seconds.


There’s no tool to help you if you can’t read aloud.

There’s a posting from a successful reader who describes how she changes her presentation slightly to make it easier to follow each character.

Yet another reason I will never read for audiobooks.


If you think about it, record a 20 second mono test and post it.

It should be your normal presentation style with no effects or corrections.

Words don’t matter—read the cereal box—and don’t make it too short.

There are some sound problems that don’t fit conveniently into the correction suite.


Thanks to all of you! Let me digest this. I’d done 4 books successfully, and two other ones were rejected. (These are all my books that I’ve written.) One indeed did have to do with that sound in the beginning and number of seconds! Except when I opened it, there were 3 seconds and the noise level was clear. I’m still messing with that. They had to see something I’m not.

This new book, though was where these problems occurred. I will work with what you all have said and let you know how it goes. Probably won’t get back to it until I’m off on the weekend. I so appreciate all the help!!


We have to divine an enormous amount of information from your initial posting, so sometimes we miss the boat a bit. It’s like speed dating over multiple timezones.

two other ones were rejected.

They usually tell you what they found wrong and suggest possible remedies. Did they this time?

The suite of tools is not a bad place to start. In English: Rumble Filter, Set Loudness (RMS), Gently trim any tips or blue wave peaks. That’s it. That’s the whole suite.

As you found, if you don’t establish a sequence or process, you can chase your tail forever—the specifications seem to be contradictory.

“Wait. If I make the RMS OK then the peaks are too loud.”

This isn’t the only published correction suite. There’s one that starts with Noise Reduction and goes from there. That’s dangerous. Noise reduction can seriously change the character of your voice—and ACX is looking for that.

Let us know when you get back. Do record that 20 second sound test.


I opened it, there were 3 seconds and the noise level was clear.

They do automate some of those tests. Magnify and look at the extreme start of the sound. It’s not unusual to hide some tiny ticks, peaks or pops up there that are hard to see if you’re looking at the whole chapter. They make it seem your presentation starts from zero instead of three seconds in.

Another possibility is “earthquake” rumble. Many home microphones produce low pitch trash that’s rough to hear. Nobody can hear it, but it throws off the measuring tools. That’s why the first step in the correction suite is a rumble filter.

I just made a quick jaunt over to the ACX pages.

Each file must have 0.5 to 1 second of room tone at its beginning and 1 to 5 seconds of room tone at its end.