Rejected by ACX

Hello, looks like I just missed KOZ while creating a username etc. I hope to catch the attention of someone as knowledgeable and kind.

I have successfully published on ACX once before and recently uploaded several titles. Unfortunately, one is being rejected for RMS issues. I have read every old post that I can find on the topic. Currently I have passed all chapters together an when I follow the advice:

Measure the peaks:
Select the whole clip by clicking just above MUTE.
Effect > Amplify. Read the top number, don’t apply the tool. It should read between 3.5db and 3db. /// I get 0/none

Measure the noise:
Drag-select between 2 and 3.5 seconds.
Analyze > Contrast: Measure Selection. It should read lower noise (higher number) than 60dB. Stunning low noise values are not automatically valuable. You never had anybody reading a story to you from a black vacuum. ACX is not impressed with extreme numbers. Unusual numbers usually cause other sound problems /// I get 76.7

Measure the Loudness (RMS)
Select the whole clip by clicking just above MUTE.
Analyze > Contrast: Measure Selection. The value range is 18dB to 23dB. This one is fuzzy and changes with the number of pauses and how theatric you are.
///I get 32.2

After Normalizing and compressing ///I get 0, 68.24, and 25.3

The entire book together is just over an hour; which 10 seconds would it be advisable to upload for sample?

Also, I have read through the forum in attempt to understand the process of pasting in room tone where you have made other corrections. Originally I misunderstood this to mean when I used Control L, but it looks like it is also necessary when noise removal has been used. I have discovered the Anylize>Silence Finder…I would be grateful to know how to correct these silences with room tone without changing the timing. I have not had success understanding and assimilating the information mentioned in other threads.

Your guidance is a gift I will use to grow and when I am able I will “pay it forward”. -rjshine

Try brownies. Most engineers will kill for brownies. Fudge is a good option.

I get 0, 68.24, and 25.3

You can’t have 0. Zero is the place on the sound meters right at overload and clipping distortion. ACX flat won’t accept that. You should reduce the peak levels to -3 or my personal favorite -3.2. That allows for MP3 encoding and accidental level shifting just before submission.

But. That will reduce your overall loudness (RMS) even further than it already is, and it is already beyond conformance.

So Something Happened. I’ve been a party to several vocal performances by now that are within spittin’ distance of conformance with very little patching and “help” to know it can be done. It’s not magic.

You can get into trouble trying to manage Room Tone. Particularly for us, you sit absolutely still and hold your breath for two seconds and then breathe and record a typical vocal presentation for 8 seconds, or longer and then cut it off later at eight seconds. A mono ten second WAV clip should fit nicely on the forum. We need that sort of Room Tone as an accessory for Noise Removal and the measurement tools. ACX needs their much longer tone as an illustration of how loud your apartment is. They don’t even want you in the same room for that one.

So record a ten second test clip like that and post it. No filters, effects, tools or patches. The Raw Clip.

I expect it to sail straight through testing and then we’ll decide what to do with the actual performance. Is the show available on a file posting service, or can it be? Say the first minute or so of the book. Before processing.

Of course, the clip might not sail straight through…


I only bake Gluten free brownies. I prefer them topped with Heath Bar crunch. Not sure if you would really accept brownies from a stranger, but I would happily send them :smiley:

I will go create the room tone clip. Will sound-cloud work for a clip of the book before edit?

One note. The clip analysis and patching stage act might not work on a much longer show. If you tire and start running out of volume about chapter 3, then chapter 3 is not going to match the rest of the show and that will throw the tools and filters off. They warn you about that in the ACX training videos. “Consistent Performance” comes up multiple times.

I thought ACX wanted a list of carefully produced chapters, not one ginormous sound file. Filtering and processing a chapter would be much better and easier than trying to manage a three hour file. As I recall, they have specific requirements for silence before and after the chapter segments.


Will sound-cloud work for a clip of the book before edit?

I think that will allow us to download the work. Just the first chapter or about a minute or so. We don’t need the whole two-inch thick book.

Let’s get through the test clip first. All sorts of problems crop up when I ask for that short clip.


I apologize for the sticky soda mouth and hungry belly if you can hear it.

I think I see what happened. You have a sharp clarity to your voice which would make getting peak blue waves and overall loudness difficult to pass in the same clip. I created a compressor setting for you, but it’s a little involved to use. I’ll write it up for you tomorrow. Time for all little girls and boys to go to bed.

Given that I can make it work with the clip, post Chapter One or about a minute of The Real Thing without processing somewhere.


Import the clip without the fan.

Select the whole clip by clicking just above MUTE.
Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2dB, OK

Then run Effect > Compressor with the settings in the attached graphic (click on the graphic).

The first Normalize is to get the work within range of the compressor. This second normalize is to get the compressed clip within range of ACX.

Select the whole clip by clicking just above MUTE.
Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2dB, OK


I expect that clip to meet ACX guidelines and it still sounds like you. 3.2, 21.5, 68.7. Peak, RMS, Noise.

Now point us to a longer selection and I want to see if the tools hold up.

Screen Shot 2015-04-14 at 22.57.14.png

How sure are you that you didn’t reverse the clips? The fan clip is slightly quieter than the no fan clip. So it appears to be irrelevant whether you have the fan running or not.


Uh oh. I don’t remember if I recorded with or without. Also, the full file is edited having used plenty of sound removal and leaving behind lots of little Silence spots. I would love to know how to patch over with room tone (acx issue). Although, another title of mine was accepted yesterday so maybe I just did something wonky and out of my regularly scheduled programming. I will delete the clip once I know you have downloaded it.

Thanks Again

Ran your process;results: 3.2, 68.6, 25.4

There is no provision to download the clip, but I captured it as a live feed. So you can pull it.

[time passes]

I’m getting very different readings from you, so something happened in the transfer. Split the clip in to a single mono track (menus on the left > Split Stereo To Mono) delete one of them, export a new WAV file and email it to me. It should be under 25MB, the limit of my mail.

Coming to a mailbox near you.


have you reviewed your email?


3.2, 23.7, 65.7 (will not pass)

Normalize > Compression > Normalize
3.2, 21.6, 68.9 (passes)


3.2, 22.7, 65.1 (passes, just)

The second clip didn’t need the compression to pass. I just boosted the volume and it fell in. Not perfectly. 22.7 loudness is right on the edge of the 23 bottom limit, but still. A performance with a lot of air in it (pauses) will drag down your loudness rating. The first clip was just too quiet. It didn’t pass and needed the compression step.

All I did with both clips was apply Normalize to the settings attached and/or Compression to the settings attached in the above order.

Given that ACX goes on and on about consistency of performance, once you use the compression, you should probably keep using it, so the total show performance doesn’t jump around.

“Anyone can sext, but if you do not know what you’re doing…”

Any reason not to use “don’t” in sentences? Is this a text purity thing? It sounds awkward in an otherwise smooth performance.

Screen Shot 2015-04-16 at 11.38.14.png
Screen Shot 2015-04-16 at 11.37.49.png

Are you submitting numbered chapters? I think I saw ACX wants it that way. Using what appears to be your normal studio technique, it should be possible to Normalize > Compression > Normalize to produce conforming chapters from the performance and then do the other production things like make the pauses beginning and after, export to MP3, etc. etc. That part is up to you.


Good Day ,

I originally turned in approx. 24 separate files with appropriate tails to ACX;6 were too quiet,9 ere too load and some chapters were on both lists.

I brought all my files open in Audacity pasted them together in one file then tried to Normalize etc.

One of the clips sent to you was not yet Normalized, the other had the process you told me in a previous post. Are you saying I should run the new suggested process after the other suggested have already been performed?

You are right about the unnatural speech, yes I was aiming to be true to the txt. Should i be applying the processes to the individual files?

I brought all my files open in Audacity pasted them together in one file then tried to Normalize etc.

That’s the question, isn’t it? Normalize (and Amplify) have no idea what content is. They blindly work on blue wave tips or peaks. The problem with processing The Whole Book is one accident, bad emphasis or noise anywhere in the three hours will destroy the whole show. That may be why the submissions failed.

I need to read it again, but I think they want you to process the chapters and then play through the whole book to make sure everything hangs together. One ragged chapter will stand out and you can repair, re-record and process it as needed.

Yes, I know. 300 chapters, have a happy day.

I was aiming to be true to the txt.

Take it up with the rights holder (author, publisher etc). My favorite audiobook author (over multiple listenings) sounds like she’s telling me a fascinating story over cups of hot tea. In colloquial English. She doesn’t use “do not.”

Two attachments. The first is the raw KOZ-KOZ clip you sent. Second is Normalized featuring the usual settings. Isn’t that what yours looks like?
Screen Shot 2015-04-16 at 12.35.31.png
Screen Shot 2015-04-16 at 12.34.56.png

I’m just now recognizing it. Did you leave off the last Normalize in KOZ-KOZ? When you went Normalize > Compress > Normalize, except you stopped after the compression step. That would fail everything but noise.


I’m re-writing my testing paper so it’s more universal than it is now. I nailed the test to a particular sound file and if you don’t happen to have that file, the process will be useless to you.

Also, that’s not the only way to do testing. There is test software I can download and use as well, but I wanted to do it using just what Audacity had available with no further fuss. Being compulsive, I wanted to be able to contrast and compare the two different techniques to make sure we were getting the same numbers. Of course, then there’s the matter of ACX Automated Testing also getting the same numbers. Goodness knows how they’re doing it.


I think I followed your directions…um, now when I follow your directions with the entire book in the DAW I end up with 3.2, 67.6, 25.4

I don’t understand your final post.