Meta Normalizer - question on functionality

I’m using the meta normalizer to equalize two clips in level… one is mastered hot (like -7 LUFS). The other is very quiet and unmastered (like -20 LUFS). The meta normalizer is set as:

  1. Match loudest clip

  2. Reference loudness of entire clip

  3. Limit digital peaks: -1dbFS

    Nothing else is checked. What I want this process to do is to set both clips the loudest they could possibly be, with -1db being the absolutely loudest peak that ever occurs in either clip (so the unmastered clip would hit -1dbFS peak, and the mastered clip would probably only hit peaks of like -8dbFS), and both clips would be the same LUFS/RMS. Instead, it’s turning down the loud clip by 1db, so loudness is -8 LUFS and peaks -.99dbFS, and turning up the quiet clip to about -14 LUFS and peaks at -1dbFS. The “loudness” has not been matched.

    How do I set it, so that it puts BOTH clips at the loudest possible MATCHING LUFS (with the loudest peak level that ever occurs in either clip being -1dbFS)?


1 Like

I figured it out… HOWEVER, I think maybe the meta normalizer needs an update. This seems unnecessarily complicated. My intuition tells me what I tried above should have worked, but it did not. I was able to achieve what I wanted as follows:

In the clips section, it needs to be set as:

  1. Match loudest clip
  2. Reference loudness of entire clip
  3. Ignore peaks

The master section output needs to be checked as well with settings as follows:

  1. Do not change loudness
  2. Limit digital peaks
  3. Maximum peak: -1dbFS


The problem with doing it this way is that I’m pretty sure you’re significantly limiting the quiet track, which won’t give you a fair comparison, depending on exactly what you’re trying to do.

It seems like the “more correct” way is to turn down the loud track post-limiter so you don’t have to limit the quiet one.

The method I’ve been using is this:

  1. Use Perception AB, “pre” as first plugin on each clip that I’m working on with Perception AB “master” in a playback processing slot.
  2. Choose a playback level in Perception AB that is sufficiently quiet. -14 to -16 works fine for me. For a track that quiet, if it’s averaging around -20 and peaking close to zero, you might need to go lower. The important thing is that this level should not clip your output if you play the quiet track back at that level. It doesn’t really matter exactly what it is as long as it’s not too loud.
  3. Load references/comparisons on reference track(s) with master playback processing off, and use the metanormalizer only on the references to set them to the playback level that’s set in Perception AB without any limiting.

What that gives you is a nicely level/time-matched comparison between the raw track and your mastered track via perception AB and at least roughly level-matched references on the reference track via the ear/listen button.

The “catch” is that if the ear/listen icon in the reference track is ticked, it doesn’t matter how Perception AB Master is set, you’ll only hear the reference track.

If it’s too quiet, turn your monitors up.

The only real pitfall of turning things down digitally like this is that you’ll have a higher noise floor either from your interface for from post-WL dither. Neither of those is actually a problem in practice unless you listen crazy loud in a very quiet room. And even if, somehow, it is…it won’t be in rendered tracks; it’s only in your monitor path.

ETA: the other significant pitfall is that you have to remember to run your monitors back down when you listen to anything not set up this way.

Incorrect. No limiting occurs at all for this process. From the Wavelab manual regarding the meta normalizer:

This tool operates on gains. It does not affect the underlying audio files or use any audio compressor.

All the meta normalizer does is adjust the clip levels up and down. You can see this if you go to the “clips” tab after running the meta normalizer. Scroll the “clips tab” to the right, and notice the “pre-gain” and “post-gain” columns. The post-gain columns should have been adjusted for each clip according to the settings you ran the meta normalizer with.

THAT BEING SAID, the wording used in the meta normalizer is VERY confusing. It uses the term “Limit digital peaks” when absolutely no limiting is actually occurring. It really needs to be reworded to something that is less confusing. Obviously most people would assume the word “limiting” refers to destructive limiting like what the Peak Master plugin, or the Waves L2, or Ozone Maximizer plugins do. The meta normalizer tool does NONE of that.

Rather than saying “limit digital peaks”, it probably should say “reference digital peaks”… or maybe “consider digital peaks”.


If I use my two clips as described above, and perform the following process:

In the clips section, it needs to be set as:

Match loudest clip
Reference loudness of entire clip
Ignore peaks

The master section output needs to be checked as well with settings as follows:

Do not change loudness
Limit digital peaks
Maximum peak: -15dbFS

With the maximum peak set to -15dbFS, when I click “apply” in the meta normalizer, the waveform redraws incorrectly… showing both waveforms totally loud and maxed out, but they play back extremely quiet, so obviously the waveforms were drawn incorrectly in this case.

You are right, this would need a change. Not that it’s by design that WaveLab does not do any traditional limiting in the montage, as the process wants to be as neutral as possible (only gains).
Note that if you add your own limiter to a clip, and choose to integrate the effects to the meta normalizer (check bow at the bottom left of the dialog), then your limiting can have an effect on the result.

On another hand, note that the Meta Normalizer in the audio editor and batch processor, are doing traditional limiting when necessary, as these are “destructive” processes.



So, in that case, what you’re doing is turning it up to match and then turning it back down to not clip.

What are you trying to accomplish, then? It still seems like turning loud things down to match is the better approach. But, if you’re not doing comparisons, that might very well be totally wrong.

This is exactly the principle of using a limiter to increase loudness. The overall gain is increased but the peaks are restricted to avoid clipping.

Are you referring to the “force equal loudness” checkbox?

Where is the “meta” normalizer in the audio editor? I see a “level normalizer”, a “loudness normalizer”, and a “pan normalizer”, but no “meta normalizer”.


I meant “Loudness Normalizer”.


Ahh ok… But still, the meta normalizer may “analyze” through our FX unless this is checked, but the meta normalizer never performs “destructive” limiting itself - only gain changes.

So the “limit digital peaks”, and “limit true peaks” options are confusing. Maybe should say “consider digital peaks” and “consider true peaks”.

The net result is all clips are perfectly level matched (RMS/LUFS), no clips occur anywhere in the montage, and the max peak is set to my choosing (-1dbFS - which is what Apple Music recommends for mastering).

OK, I finally figured it out!!! Here is the correct way to set up the meta normalizer to accomplish this:

I didn’t have “force equal loudness” checked!!! This is now working and doing what I had hoped…

“restrict digital peaks” or "“confine digital peaks”, I will consult.

Even this seems to potentially “imply” a destructive process…

Maybe “reference digital peaks”, or maybe “consider digital peaks”?