Montage loudness / meta-normalizer questions

Per the help file on Montage Loudness dialog box :
When Clips is activated, the gain settings of all clips in the audio montage are adjusted individually so that all clips play back at equal loudness.

When Audio montage output is activated, the general gain setting of the audio montage is modified so that the audio montage mixdown matches a given loudness and optionally does not clip.

This seems to do an excellent job, but I’m surprised by the overall gain being added to clips when I use the “Match Loudest clips”, compared to what I had in WL7. Even at the lowest vertical zoom, most of them are getting huge boosts of up to 9dB, meaning their waveform completely fills the clip. The master section shows the resulting levels as fine, but the track meter is pretty much a solid green.

My questions are:

  1. Is there a way to scale the display so it falls within the range (so the contours of the waveform are more visible)?

  2. Is there somewhere we can see and/or adjust the general gain settings of the audio montage that this loudness process makes ( I can see the per-clip adjustments in the montage, but don’t know where the massive gain reduction is happening – clips that are normalized to around -1 dB are getting a +8 dB boost, and the master section meter is still under 0 dB, so i know this montage-wide gain reduction is happening).

  3. Am I crazy, or do the tracks in WL8 montages no longer have a gain control? I used the Montage Loudness utility to “standardize” a bunch of music clips, then added a vocal track another with some musical underbeds. I needed to tweak the gain of just the vocal track relative to the previously standardized music clips track and saw there are no longer track gain controls.

I don’t know about the other points, but the track gain control is now a slider in the effects window: p.339 in the manual.

Thank you. I searched high and low in the help but didn’t find any reference to this change. A “What’s New in WL8” section of the manual would be very welcome.

As far as the other questions, when I changed the Meta-normalizer Clip options to from “match loudest clip” to "match specific loudness (-23LU), the clip adjustments indeed went to my familiar WL7 all negative (attenuation) instead of large per-clip boosts.

If I understand it right, the end audio result is going to be the equivalent if the overall limiting value is the same and the Audio Montage Output stage settings match (-23LU):

“Loudest clip” scenario boosts all clips to match the loudest track, then has to reduce overall montage gain (because of the limiter value and the Audio Montage Output Stage) DOWN TO -23LU value, while the “match specific loudness” is cutting all the clip gains to equalize them, but then boosting everything in the Audio Montage Output Stage UP TO the -23LU value (match specific loudness), subject to my 0 dB peak limiter setting.

If the above paragraph is the confirmed to be the case, (i.e. the Audio Montage Output Stage settings “correct” everything back to the same final audio result) , I will stick with the specific loudness option because I prefer the visual results of the clip waveforms.

Care: you must think about average loudness, not about peak levels. The level at the output of the montage, is used to change the average loudness of the whole montage.
You could have a very quiet clip, and a very loud clip. That could give a certain loudness of the whole montage. And the same loudness could be achieve with 2 clips with a medium loudness.

OK. I think having both the clip and the audio montage set to the -23LU would be my best option for creating a consistent level from show to show. The music files are commercially-mastered high quality mp3’s or .wav’s so should mostly fall within a certain range. Thanks again.

Wait, so then how do we get peak normalization of the entire montage? In limited playing with it in v8 so far, I found this all somewhat less-straightforward than I was expecting (could be me). It should be very clear in each case whether it is doing peak, RMS, or other normalizing. I have usually opted to adjust relative loudness in each clip by ear with clip gains, but still need a way to peak normalize the entire montage…

I’ve kind of cobbled together my own workflow for this, but not sure it is the best or most efficient. I’m producing a daily show for radio broadcast combining 4 tracks in wavelab - Music, Vocals, Music Underbeds, and Station ID’s

I typically used to metanormalize the music track so all tracks were about the same level (-1dB), then brickwall limit / compress and normalize the complete vocal .wav file to get it to a fairly consistent level (-2dB) before chopping it up into the sections to place in the Vocal track in the audio montage, then I would metanormalize the underbeds to a lower level - (-16dB) so they are a consistent level I could simply drop those into the Underbed track aligned with the vocal clips.

I then put a brickwall limiter on the master output and rendered. Now that the loudness mastering tools in WL8 are more sophisticated, I’m wondering if I can consolidate or eliminate some of these steps. I’m thinking it might be possible to run the Loudness Normalizer on the whole montage at the end using the -23LU defaults and just use the track gain to change the relative levels of the underbeds vs. actual music/vox. I too am still confused by some of it i.e. not sure if the Montage Loudness normalizer would counteract/undo the track gains if they were set before loudness normalization (if I reduced the gain of underbed track BEFORE running the global loudness normalization, would that track get boosted back up to full loudness/peak levels, or would it respect the relative track gains?) I will have to keep experimenting. I would be interested in hearing about other people’s workflows or pointers toward any tutorials in this area, as I have been doing it by trial and error for some time.

Wait, so then how do we get peak normalization of the entire montage?

The Loudness Meta Normalizer is not designed to peak normalize the audio montage, as it deals with loudness.
However, you can use it to achieve peak normalizing in the following way:

  1. Set an high gain like this:
    2013-05-23_22-17-52.png
  2. Set like this:
    2013-05-23_22-14-41.png
    This will result in adjusting the gain slider (#1) according to the specifoed peak level.

I have usually opted to adjust relative loudness in each clip by ear with clip gains

The Loudness Meta Normalizer can do this for you very well, and very quickly…

I too am still confused by some of it i.e. not sure if the Montage Loudness normalizer would counteract/undo the track gains

The Meta Normalizer change the clip gains or/and the Mastergain, but not the track gains (the lanes).

Whoa, huh? Well, thanks for the clarification…But isn’t this incredibly convoluted for such a essential but crucial operation as peak normalizing? This seems like exactly the kind of thing I’m expecting WL to do absolutely easily and transparently.

The process you suggested does seem to work, but it needs to be less of a workaround and to operate at the track level. The normalization of the master also takes into account where track gains are set-- This makes sense, but doesn’t work well when this process needs to be done to multiple tracks.

A pretty basic workflow would be to edit several takes of a single piece in the montage, make any clip gain adjustments for edits within the piece, then do a simple peak normalization of the entire montage. This seems fairly straightforward- what am I missing that’s making it so involved?

Really, all the RMS normalizing and loudness-matching functionality is nifty and I’m sure will be useful for me at some point. But for many of my uses it’s a big leap of faith right now, and would absolutely need to be checked by-ear anyway, so it’s not a panacea and is lower on the list for me.

Didn’t WL 7 do a straightforward peak normalize? How is this designed to be done, PG?

Thanks again,
Brian

The concept is: “loudness is the important parameter, peak level comes second”. Simply because end listeners don’t care about peaks (abstract to them), but about loudness (what they ear).
Hence the tool is oriented around loudness normalization, not peak normalization. But peak limitation is necessary with this concept. However, as explained, there is still a trick to do it, if needed.

Right, I understand (and agree) with the concept; I just expect a professional editor to allow me to handle both peak and RMS normalization in either way (automated or manual). Simple peak normalization is a useful feature and one almost every free editor provides. So please consider this a feature request to add peak normalize to the montage with the same options as RMS has.

Thanks,
brian

Sorry I’m so late to this thread, but while loudness matching already loud clips (0.2 headroom) using the Meta Normalizer in Wavelab 7 and Wavelab 8, I noticed that Wavelab 7 reduced the level of clips relative to the loudest clip if 0 peak input to MS is specified (which is what I would expect).

But Wavelab 8 (if using the factory default preset)(as said in this thread) raises some of the clips well above zero to do the loudness matching and then reduces the overall montage level (which I can see on the Effects tab under Master). It also makes all the clips into a brick (not helpful visually).

It’s more than a little disconcerting to see clips raised well above zero which would blaze into the red if not for the overall montage volume being reduced. Maybe it’s technically correct, but many engineers would have issues working this way.

Can Wavelab 8 be updated to do it as in Wavelab 7 (reducing clip volume relatively to loudest clip, and maximizing peak)? That made so much sense.

In Wavelab 8, if only the Clips checkbox is checked (“Montage” not checked), with “match loudest clip” and 0 digital peak selected in the Clips section, with program already loud (0.2 headroom), the process only seems to Peak normalize, not loudness match, when it should loudness match by reducing clips relatively with peak at zero as Wavelab 7 did. That seems like a bug, because the same settings WILL loudness match if you take those 0.2 headroom clips and lower them all 6db with the volume envelopes. But not if you leave them at the original -0.2 level. And if you do lower them all and loudness normalize, how do you get them all back to 0 peak?

You can try this with any commercial rock release that’s already loud (like probably the AC/DC Wavelab 7 example on YouTube) and it won’t loudness normalize in Wavelab 8 if using only the Clips section to match loudest clip and limit to 0 peak. That worked in Wavelab 7, and made sense. Is there no way to put the Clips section as Wavelab 7 was, and then gain stage from there? Also with Wavelab 7 you could easily Peak normalize only the entire montage, which many people find very useful.

I am not aware of any bug in the WaveLab 8.0.2 meta normalizer (it was much tested). I also think it’s better than WaveLab 7 implementation.
Your description is a bit complex on paper, hence if you can provide a short test montage that shows an abnormal result according to you, I can analyse what happens.

Thanks PG, I think it’s better than WL7 too, with all the LU, True Peak, and Montage and Master Section options. Maybe I shouldn’t worry about the “well over zero” level thing with the clips if the system can handle it, but it does seem very unusual, and visually it’s a drag. Also a simple Peak normalize of the whole montage seems pretty standard in other programs.
I’ll try to make some pictures tomorrow night, but basically the problem I see with the Clips section can probably be replicated by loading any loud commercial CD, selecting Clips checkbox only, Match Loudest Clip, Reference: Loudness of entire clip, Limit Digital Peaks, Maximum Peak 0.00. With these settings, the results will change the clip levels by maybe 0.1 or 0.2 db (not expected, and average levels are not matched). If you then turn all the clips down with the volume envelopes by say 6db and re-run the same thing, the result will probably change some clips by 1 or 2 db because it’s now volume matching, as expected. I’ll get some pictures tomorrow night.
With the equivalent settings in Wavelab 7 it works correctly, even on the loud original, by turning down the clips that are too loud.

I must agree here that peak normalization IS needed. Every software that converts floating point audio (large number space) to fixed point (small number space) should have it to maintain the maximum signal integrity and highest playback levels.
WL 3 - 7 did this very easily with the meta normalizer. Now WL8 can only reduce levels to come to a specific peak level in one step (yes, I’m aware of the workaround above). It takes care of clipping, but not of too low levels on a CD.

I will probably never understand why this specific function has been cut down in the the meta normalizer. Maybe this is something like if you ask in an internet forum how to peak normalize, you will only get answers like “normalization is bad, you do not need it, use compression”. Ok, when producing pop music the compression will “kind of” normalize for you, but what, if you do not want to change the dynamic range of the audio signals? You HAVE TO normalize to get proper levels on a CD.
I know it is hard to believe, but there is music out there, where the tracks must not have the same loudness and the relative loudness does not matter at all. Here, the music defines the dynamic range of a recording, not a broadcast standard.

Nils

The meta-normalizer never compresses the signal, hence never changes the dynamic range of the signal. It always normalizes the peak level, but normalizing is not the same thing as maximizing. It normalizes the peaks so as to reach a certain loudness level. This is why maximizing happens if you choose to have a high loudness level.

Priority is given to loudness against peak level in this tool. As already mentionned, I will add a “maximizing” option in a future update, for people who don’t care about loudness balancing across an album, or don’t care about loudness references in general.

Yes, sure. I am aware of that.

Thank you for not forgetting this :smiley: I am really looking forward to see this feature again in the future!
Wavelab is still the best!!!

Nils

PG, I’ve finally got these screenshots for you. First off, if someone uses the Factory presets, they’re not going to have a problem with volume matching or peak levels, like I’m about to describe. Those are handled by level offsets at the montage or master section in the Factory presets:
Wavelab 8 Before MetaNorm.PNG
Wavelab 8 Settings 1 (DefWDig).PNG
Wavelab 8 Settings 1 Result (DefWDig).PNG
The issue I see is if you use the following settings:

Clips section checked (Montage and Master Section unchecked).
Match Loudest Clip
Reference: Loudness of entire clip
Limit Digital Peaks
Maximum Peak 0.00

My screenshot of those settings is called “Wavelab 8 Settings 2.png”. (in the following post).


This is a normal / loud pop rock album loaded from a commercial CD, which peaks near zero. The original is shown in the shot “Wavelab 8 before MetaNorm.png” (above). The result after Meta Normalizer (with my “clip only” settings - settings 2) is shown in “Wavelab 8 Settings 2 Result.png”. (in the following post).

There is virtually no volume matching going on in that result, although there should be. Apparently that’s because Wavelab 8 only goes up from the loudest clip to match volumes (as seen in the factory default “Wavelab 8 Settings 1 Result.png” above). But because the “Clips only” setting I want to use includes a zero peak ceiling, Wavelab won’t volume match the clips of this normal / loud recording because I’ve told it not to go over zero, and they’re all already very close to that.

If you bring all of the clip volumes down 1db (with the volume envelopes) and re-run the Meta Norm, it “starts” to do volume matching (but only partially, only what it has headroom for). If you bring the clips all down 2db, it does more volume matching. Down 3db, even more. Until it finally has room to entirely volume match under zero. This just doesn’t seem right, because the ability to do any volume matching at all is dependent on the original levels of the clips, so it’s probably going to be wrong in almost all cases of normal / loud clips.

Wavelab 7 took care of this by reducing compared to the loudest clip (not raising) if a zero ceiling was set. It would put the entire montage peak at zero, and lower all clips relative to the loudest clip. (see Wavelab 7 png’s in the following post).

Is there a technical reason not to do the Clips section the Wavelab 7 way? It makes more sense visually, for one thing. Also, it seems safer and would probably be more familiar to most engineers who would not be inclined to raise those clip levels well above zero when they’re near the top already.

Regardless, the Clips section settings I’ve described are not operating correctly on all program material, and should be fixed IMO.

(Sorry, split due to limit per post)
Wavelab 8 Settings 2.PNG
Wavelab 8 Settings 2 Result.PNG