WAV File passes ACXCheck, MP3 doesn't. Wondering why?

I have exported the exact same file to a WAV file, and an MP3 file.

The WAV passes the ACXCheck

The MP3 file doesn’t.

Is this a function of the MP3 losing some … quality? or something else.

Just wondering.

That’s why we recommend checking the MP3 before submission. Try encoding the MP3 with a higher quality. The original WAV is right on the edge of passing. The damage caused by MP3 compression pushed it over.

ACX has minimum MP3 quality specifications and maximum file size requirements and both of those numbers are generous. Given what they’re going to do to it, bumping up the MP3 quality is recommended.


The WAV peak value is right on the ACX specification of -3 and the RMS (loudness) value is only 0.2dB away. Any MP3 compression error might put you in trouble.

Try this:

Audio Compressor
– Select the whole clip or show by clicking just above MUTE.
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2 > OK
– Effect > Compressor: Thresh -20, Floor -40, Ratio 2:1, Attack 0.2, Release 1.0, > OK
– Effect > Normalize: [X]Remove DC, [X]Normalize to -3.2 > OK

That should gently push the loudness up and put all three values into passing. This is a normal error, by the way. Nobody ever walks away with a high RMS unprocessed show.


Thanks Koz,
It’s an interesting business, no doubt.
I also discovered today that if I use the Master file. That is, the one actually created in Audacity, and get it right, passing the ACXCheck, if I export it to MP3 right there, then load the MP3 back into another instance of Audacity it passes when I check it. If I Export the master file to a WAV file, load the WAV back into another instance and apply the effects to that, it passes the ACX, then export that to a MP#, load the MP3 back in, it usually don’t pass. Not by much, but still.
So obviously the thing to do is do all the edits, and effects, while still in the recording session, on a copied track, with the original muted… it all seems to work a little better.
But anyway, I take your advice. Most excellent words of wisdom, thank you. This learning curve I’m on is a doozie!

I seem to have discovered too - and it’s probably not the thing to do … but where I’m close to passing the ACX, usually the peaks are too high, if I apply the Limiter, or Compression then Limiter, and that’s all, it pulls everything into line, and usually passes the checks? Amazing.

However, I will print out your advice - the 3 steps to audio heaven … :slight_smile: and tack them to the wall by my head! Thank you for that. Extremely useful.


ACX Specifications are also broadcast. Anybody with a good quality , quiet performance can do it. I did a test clip in my third bedroom just to prove it can be done. Adjust the volume and out the door. Much like your submission.

However neither makes it by much. They’re brittle. If the humidity goes up, it can fail ACX. Usually, gentle corrections are enough.

That’s far better than creating trash and trying to “rescue” it. ACX has traps for people trying to do that.


3 steps…

Which three steps might those be?

Recording audiobooks has resisted easy formulae.


Looking at the figures in your first post, a little bit of dynamic range compression should bring the original recording comfortably into the specified RMS range. The MP3 will then also probably be within range.

The Noise floor looks suspiciously low, like you may have used too much noise reduction. Listen critically to the exported MP3 on good headphones - does it sound “processed”? It shouldn’t - it should sound “natural” as if it has not been processed at all.

Bad noise reduction can sound like hissy words or words with noise tails or “leading” tails. That’s if it didn’t reduce enough. If it reduced too much, you can have honky, wine glass voices. I don’t recommend 18, 6, 6 any more because it always produces borderline or unacceptable works. Anybody can tell just by listening there’s something wrong and the works certainly won’t pass ACX Human Quality Control.