Not Passing ACX Check Analyzer Issue #2

Audacity 2.1.2
Windows 10, 64 bit
Not Passing ACX Check Analyzer. I need help with Effects

Audio Software: Audacity 2.1.2
Audio Interface: Scarlett 2i2
High Pass Filter: Shure A15HP
Mic: Blue Spark

I posted an issue a couple of weeks ago With similar issue. I have set for hours trying to get the Peak level up and RMS to comply with ACX.
ACX requires the file measure between -23dB and -18dB RMS and have -3dB peak values and a maximum -60dB noise floor.

I would sincerely appreciate knowing the mastering that passes ACX.

Thanks again.
(2)ACX Pk -3.2 RMS -17.1.png
(1)ACX Pk-6.0 RMS -19.9 .png

ACX requires the file measure between -23dB and -18dB RMS and have -3dB peak values and a maximum -60dB noise floor.

That’s not quite correct. Yes, you have to hit RMS (Loudness) between -18dB and -23dB, the other two are open-ended extremes.

Noise no louder than -60dB when you stop talking and no peak louder than -3dB—ever. ACX-Check is correct.

Are you using SetRMS? Depending on how old these posting are, we changed the recommended process. We never had a good way to force RMS values where we wanted them. Steve wrote a tool that does that, so the process is a good deal shorter than it used to be.

The golden chalice is to read in a very quiet, echo-free room with a terrific microphone and not need any processing past overall volume. I can do that and we recently had a poster who could do that. Live reading will always need a little volume adjustment because the specifications for reading and posting are different.

If you’re not using SetRMS, it’s still possible to hit it with the other older tools, it may just take a bit longer.

I need to drop out for a bit. It may be easier if you post a sample of your work on the forum.

You may not have to change anything since you’ve been shooting for the wrong standard.


Thanks Koz for your willingness to assist.

I have tried to follow your instructions in setting up SetRMS. I still can’t meet the ACX Requirements.

I don’t see the file I uploaded.

It’s not quite that easy. I’m working in the field right this second, and I think I’m missing one of my software tools… !@#$%.

You have a distortion problem on top of ACX-Check conformance troubles.

Nice voice, though.

As we go.

I need to write a summary of this instead of dumping you straight into the comments, but this is the AudioBook Mastering document I wrote. Chew on that while I get back home. It’s in blocks, so scroll down.


Thanks again…I’m getting ready to watch some basketball…be out the rest of the night.

I look forward to ruminating on your Mastering file.


The submitted clip is AudioBook compliant just as it is. But there is something wrong with it and I suspect the AudioBook people will flag it right away. See where none of the blue waves ever get taller than 0.5 up and down?
Screen Shot 2017-02-04 at 17.25.59.png
That’s sound distortion. If you zoom into the waves, you’ll see the tops and bottoms get cut off. The waves toward the right in this pix are normal, but ones on the left have distortion. That’s called clipping for what the waves look like.
Screen Shot 2017-02-04 at 17.24.21.png
If you do that enough, the voice starts sounding harsh, gritty and crunchy. ACX will bounce that in a heartbeat. That’s one of the reasons your Peak volume numbers are so low. They pass, but they’re lower volume than they’re supposed to be because of the distortion.

Didn’t we go through this once?


While Koz hikes in from right field I pulled down the sample and had a listen.

Measurement wise it passes the ACX requirements with flying colors.

I would almost say it’s too quiet – did you use one of the noise reduction tools? I don’t hear any side effects, but the noise floor is quite good.

But there is something in your process that is clipping. Particularly noticeable in the test clip at “God”. So so the first question is: Where are the peaks in the original recording. If they were at “0 db” then it is just a matter that you need to back off the gain knob on your Scarlett. When recording it’s good to keep one eye on the recording level meters, occasionally hitting the yellow is good, hitting the red is bad.

If the peaks in the raw recording are at the same -6dB or lower then not only do you need to back off the gain knob on the Scarlett, there probably is a gain knob in the windows control panel that you need to turn up to 100%.

We’re baaaaack.

Your test for this overload problem is yell (Do Not Blow) into the microphone.


A normal microphone system will create tall blue waves all the way from 1.0 up to 1.0 down. The sound meters should bounce all the way to the right and turn red. You are intentionally stressing the system just as a test. It doesn’t cause permanent harm.

I bet yours doesn’t do that. I bet yours creates a solid block of blue that never gets any taller than 0.5. That’s wrong and it can harm your show.


Does that problem go away if you record in stereo? Give that a shot. You may be one of those lucky people that has to record in stereo, delete one track and convert to mono later.

If that’s it, then that reduces us down to straight voice quality and ACX issues.


It is as you said…I can’t get above 0.5
I tried stereo…same results get above 0.5
I looked in the Device Manager for Audio inputs and outputs haven’t found anything yet even by adjusting some gains.

Where do I go from here?


Run through your sound channel again. Scarlett 2i2s come in first and second generation. There are two different Spark microphones.

When you’re doing your woof test, do the lights on 2i2 knobs turn red? I think they’re supposed to if the system is close to overloading.

I expect you to have two blue waves in Audacity if you’re recording stereo and only one of them has your voice on it. The other should be flat. Is that what you have?

We could be dancing down the garden path with this, but overloading at 0.5 is a thing and it’s different from something just running out of steam at some random loudness value. There are specific failures that cause damage at 0.5.

There is a desperation method. If you absolutely have to get a voice track out the door, you record it at -12 and -6 instead of -6 and 0. You record the whole thing low to avoid the distortion and them bump it up to normal volumes in post production. Everywhere we say “occasional peaks at -6”, you record it with occasional peaks at -12. You have a well-behaved system and can probably get away with that even though it’s not perfect. If there’s somebody with the car running in the parking lot for your track, this does work.


Does Audacity present you with an input volume adjustment? If so is it all the way up?

After that my next thought would be to look for drivers on the Focusrite site website.

Koz…I know what the problem is…
When I started this audio book project, I changed to a new computer. This morning I hooked everything back up to the Old computer and Peaks would spike up to 0.dB no problems.

I have tried to match the system audio files from the Old computer to the New. Thus far, I don’t see the difference. I’m going to join the never ending circle to contact the new computer Co. (Lenovo) Tech Dept. today. I trust they quickly know how to fix the problem LOL. (I’ll raise my faith levels by the time I make the call.)

From a technical standpoint, what would be the best possible statement I could make that would pinpoint the problem to them?

MegaThanks for your direction. I’m contacting the Author, (who is extremely understanding) and if I can’t resolve the issue with Lenovo, I’ll start re-recording the 27 Chapters. The good news is, I love the book and the author and look forward to starting the project over if need be.

I wouldn’t bother with Lenovo. Chances are excellent this is not a computer problem, even though that’s how it looks.

That and the split nature of the problems(s) makes this challenging.

Let’s do “out the door” mode.

Let’s say you’re going to read on the old machine until you get the new one sorted. Read a 20-second forum test and post it. I’m betting with minimum fuss, we can make a processing suite and you’re done.


A 20-second forum test from the old machine. This file has already encoded. Are you wanting a file before encoding?


Are you wanting a file before encoding?

Before you do anything to it. That’s why this request usually has to be shot apart from other, older works.

You should post clean, unprocessed work. Don’t adjust > anything > before Export. No filters, effects, adjustments, tuning or processing unless otherwise told. >

That’s a direct quote from this document.

We can’t take processing out of a work, and having multiple people offering processing advice is just asking for trouble. That and MP3 conversion has been known to mess up the peak number. I built that possible error into the newer tools.


This is where you get busted for a suspiciously quiet voice shoot. Remember several forum elves commended on that.


Ok…bust it! I’m ready. This was recorded with High Pass Filter.

This was recorded with High Pass Filter.

And it was recorded on the new computer. It has the 0.5 distortion and damage on the blue waves.
Screen Shot 2017-02-06 at 16.25.19.png
We can’t solve clipping distortion. Effect > Clip Fix is only good for very brief distortions and even then, it only guesses at the repair. It doesn’t actually know how to fix the waves. We can’t fix whole words that get damaged.

Record a new clip on the old computer so it doesn’t have any distortion and post a WAV, not an MP3.

Start the recording, pause quietly for two seconds and then read something or talk for 18 seconds.