Loudness Normalization - Album Normalize vs Track Normalize

WaveLab
#1

All,

A question today on using WL 9.5 and it’s loudness normalization. I have been doing some tests with 9.5 on groups of songs that I use to make mix CDs.

As many have undoubtedly noticed - making a mix CD using tracks from all different eras can result in a mix that makes the user have to constantly work the volume knob either up or down depending on what’s playing.

I am using WL to normalize my mix tracks to a consistent value (like -14LUFS for example) to ensure that any mix I do - has all tracks in the same volume ballpark.

But the one flaw with this approach is most likely based on the settings I am currently using within either the batch Loudness Normalizer plugin or the Meta Normalizer within the Montage.

If I include a couple of softer songs in a mix - I do not necessarily want them hard blasted up to a constant -14LUFS - rather I would like them to be normalized using the loudest track of the group being analyzed - rather than assigning a hard value.

I read an article today about how Tidal is changing their normalization processes to “album normalization” rather than “track normalization” and want to know what settings could be applied within WL to accomplish this for my mix CDs.

Give this a read and let me know if you have any suggestions:

http://productionadvice.co.uk/tidal-normalization-upgrade/?awt_l=P5h8s&awt_m=3ZvsqOtN9eNU.YV

Cheers,

VP

#2

I don’t think a solution can be found in WaveLab because what Album Normalization does is measure all the songs, set the loudest song to the desired level, and then adjust all the other songs by the same amount (not to the same level) so they levels from song to song stays relevant to how the original mastering engineer intended.

Have you tried the “Top Of Loudness Range” setting? I know that setting is in the Meta Normalizer in the montage, not sure about the batch processor. To my ears it’s more natural/musical.

Interestingly though, I did just request a way to normalize 1 clip, and then have the gain of all other clips be changed by the same amount to keep the relative levels intact.

But back to the question, I don’t think there is a way because Album Normalization is meant for albums with songs originally mastered together to keep their relative levels intact, and Track Normalization is meant to get all songs the same measured loudness. Again, the “Top Of Loudness Range” setting might get you closer than normal normalization.

#3

My spin on this … for what it’s worth … is that you really need to manually adjust levels/master the individual songs separately to achieve the outcome of relative vocal levels over different program material. It’s not really something that can be “automated” in any DAW as taste is a factor that psycho-acoustical solutions can’t offer.

#4

I hope to add this feature sometime. This being said, you have a semi-automatic way of doing it now, and this relates to Rat’s suggestion.

  • Use the meta normalizer to the single clip you want to adjust to a given loudness.
  • Look at the log and see how much gain was applied.
  • Manually set that gain to the other clips you want to “follow”. This is the tedious part.

Let’s say the log indicates a gain of +2.3 dB for clip A.
Then for clip B with has no gain, you will need to set +2.3 dB
Then for clip C with has already a gain of -1.2, you will need to set (-1.2 +2.3) = +1.1 dB

#5

Thanks PG. The tedious part is what I would hope to avoid. I wonder if it can be done with scripting but certainly, a check box in the Meta Normalize Window would be a nice easy solution for the user.

#6

+1

#7

Tedious is right! :slight_smile: I am not sure I am ready to manually adjust stacks of tracks to make a mix CD. However - since you already have the making of algorithm there - it would be nice to have WL do the work for us in a future update.

Side note - that Tidal article - I believe - is just the start of this “album normalization” trend. I can see this logic easily spilling over to the other players like Spotify, Apple Music etc. Would be fantastic to have the ability to normalize a group of songs against the “target” - right inside WL.

Cheers,

VP

#8

Or you could select all clips except clip A, hold down alt, and raise all the volume envelopes 2.3 db at once.

But this assumes all the tracks were at the “correct” relative volume in the first place. Album normalization just raises or lowers all of the tracks in the album by the same amount. I agree with Rat, it would be a hard thing to automate perception after the fact, if a couple tracks seem too loud in a mix of lufs integrated normalized tracks.

I’ll have to try the top of range as Justin suggested.

#9

I assume with mix CD you mean you line up tracks in DAW with some overlaps, or use DJ software like TRAKTOR, or use records or CDJ’s correct?
why would you use the loudest as a reference? why not the least loud? in all these scenario’s you can (and need) to work the mixer faders to make this mix work in a unified way. That’s the DJ’s job :slight_smile: Not sure why you want to inflate older tracks and ruin their dynamic range to cater for modern tracks.
mix to target loudness by just mixing the loudest tracks less loud. You can always add a limiter over the recorded mix to unify levels even more.
just saying, this “problem” should already be 99% fixed before polishing the complete mix in WL.
If you really want to go the other way around, I would loudness normalize all tracks to the least loud. so you preserve all transients, rather than limit the heck out of older tracks, skewing their dynamic relationships

#10

Correct. I am using Studio One to assemble my CD mixes - intended strictly for use in the car.

I did not indicate a need to necessarily use the loudest track as a reference - but for a noisy environment like a car - a louder target - is much better than a quiter one. Ideally I want to be able to set the target reference for the grouping - but have them normalized against a specific target to achieve a loudness “balance” across the entire group.

Right now all I can do is set my loudness to say -13LUFS and have WL “move” each track to be inside this target range. This truly does ruin the dynamics in some softer songs - and is not realistic over a given mix.

Understood. But then again - these mixes are not designed to be audiophile friendly nor are we looking to necessarily preserve transients either. If I have a mix with a track with a LUFS of -7 and bunch of songs at LUFS -14 - the idea is to find some common ground (LUFS -11) and bring the loud down a bit and the lighter stuff up a bit to make the mix relevant.

The Tidal approach seems to not assign a ballpark LUFS value to each track within an albums worth of songs - rather it attempts to “even out” the loudness experience by raising the loudness of the overall set of tracks by a specified (or maybe even an unspecified) amount based on whatever algorithm they are using to achieve this.

When I read the article - it makes sense to me - and how PG laid out his idea - is pretty much what I think the Tidal process is doing. Would be nice to see this in WL some day.

VP

#11

One of the issues with these level matters is that they are a “moving target”. Psycoacoustic processors may be getting smarter, but reconciling relative levels for a track or style that has the vocal mixed back or a searing high energy guitar solo will likely present normalization challenges. Typically, you will need the intervention of a carbon based life-form in order to acheive satisfactory, professional outcomes… If you have ever tried to mix to a broadcast deliverable, for example against a dialnorm, you will know what I mean by this. I think it was Bell that did a lot of research into this sort of thing a number of years ago.

#12

You have to do balancing before Tidal does their Album Normalization because that’s the whole point. They leave the album tracks balanced exactly as the mastering engineer balanced them, but with the loudest track at the lufs target (as opposed to Apple using the avg level of the album at the lufs target). But they’re not touching the track relative balances they’ve been given. So it’s not going to work on a bunch of unrelated tracks that have not been balanced beforehand. As far as I can tell, PG was illustrating the Album Normalization part only, so you still have the question of how to balance the unrelated tracks automatically in the first place, so that the 2 tracks that are perceptually too loud after normalization are reduced automatically somehow. I think Top of Loudness Range as Justin suggested is still something to try for that.

#13

Agreed. It’s also possible that I am overthinking this a bit too. I did a couple of test mixes where I just let WL apply a -13.5 LUFS loudness target to 20 or so tracks. When I lined them up for a basic CD mix and did some basic fades etc - the whole package sounded great.

I think making sure that the loudness “pocket” is not all over the place (Resulting in excessive volume knob twisting) is much more important to me then worrying about whether all tracks are “balanced” against the loudest one of the bunch.

VP