Normalise or Not Normalise?

[michael11]

That's it really!

I was wondering what the current thinking of the members is on this subject.

Michael.

[Russell Landwehr]

Heya Michael.

After this thread http://forums.taxi.com/topic125811.html
My position on normalization is: do it to a mixed track but not to individual tracks before mixing. BUT try to get as close to the target peak as possible when mixing.

Very seldom to I need to normalize since I'm setting a limiter/maximizer (Waves L2 or L3) on the master bus near the end of the mix process. This gets my mixed cue hitting my -0.5db target. I have had a couple of delicate cues lately, though, that when mixed were hitting about -1.0db or slightly less. I normalized those cues in my editor.

Russell

[andygabrys]

good call Russell. I forgot about that.

[Kolstad]

I often normalize individual tracks to allign peak levels before mixing (-6db or -10db). It's not neccessary at all, but I like to do it.
More important is the percieved loudness of the track, though. Sometimes a track can sound louder at -10db than one at -3db. In some of those cases I don't normalize, so it's not a rool, just a tool.

I've never normalized a finished mix. What does that do?

[andygabrys]

Magne,

normalization of a mix is the same as just inserting a gain plugin in the master chain, or turning up the master fader, or raising the level of every channel that feeds the master by the same amount

you are just making the mix louder with the same balance.

say you mix with about 6 db headroom (peaking around -6 db).

Then you add some buss compression and are not careful to make the plugin in/out unity gain, so it adds a couple db of level.

then you add some master eq and it ads another couple db of level.

so maybe your mix is now peaking at -2 db.

when you start in with your limiter, you are going to have to add 2 db of gain on the input before the limiter starts chopping off the peaks. Some limiters have an input makeup gain that works in just this way, and then a secondary gain control to squash it once you have got the input close to 0.0 db. So this scenario you don't need to normalize.

but then imagine another scenario where you are making a compilation CD, and you got some tracks that were mixed with 10 db or more of headroom. And your tracks are all about 6 db headroom. You might want to normalize the 10 db tracks so that they peak at the same place yours do, i.e. normalize to peak level of -6db. It just means that you will using the same operating ranges on your compressors and limiters when you go to master.

That's all it is, just getting things in a similar zone before you start - with the caveat that of course just adding gain will also add volume to the noise level, so if you have the choice, doing it in the mix instead of the stereo mix file would be preferable, if you start to notice the noise and hiss.

[Kolstad]

Tnx, Andy. That makes sense.
The principles involved when normalizing a mix is somewhat similar to normalizing individual tracks then, it seems.
Gotta watch the noise floor, consider the headroom, and the percieved loudness too.

[Russell Landwehr]

Gotta watch the noise floor, consider the headroom, and the percieved loudness too.

yes, those are important, that is why I mix with my peak target in mind and my mastering plugins active. That way, all that stuff is part of the mix-down and I don't have to go back and re-mix after normalizing in an editor. However, I am mixing tracks for one-off cues and not part of an album. I wonder if that would be detrimental and screw up the mastering process for an album...

When I have tracks that are too quiet, I don't normalize it before mixing, I just turn up the gain or make-up gain in whatever plug-in I'm using on that track. And if there isn't a plug-in on that track, then I insert a simple non-CPU-hogging plug-in so I can turn up the gain on that.

Russell

[elser]

I've never understood the down side of normalizing. Sure your raising the noise floor but you also do that with compression/limiting and you also do that by pushing the gain. So whether you normalize on individual tracks or on the final mix what do you lose by normalizing? It seems to to me to be an easy automated process to get a track to its optimum peaks.

[andygabrys]

compression / limiting actually introduces a plastic non-reversible deformation of the audio. you are changing the ADSR envelope of a sound. Which in many cases is desirable cause its the sound we are used to.

way different than just changing the volume.

and ultimately when it comes to volume change, normalizing, pushing a fader up or whatever, it only matters if the result is worse than what you started with.

[Len911]

I would think the only downside is if you unintentionally decreased your headroom if you normalized each track. Normalizing is useful in batch processing after every processor to ensure you don't end up with distortion in wavelab for example.

I have in the past used wavelab's loudness normalizer. I've only used it say after I've mastered a track and in a player the volume has to be maxed out or if after I've posted to soundcloud the volume has to be max, I like it about midway for a comfortable listen.
In wavelab I can check the loudness and then try out different levels, for example, if a track is averaging -24db and it sounds too soft, I can try -15db, and maybe it sounds too loud, I can undo the loudness normalizer and try say -19db, and repeat the process until I find the sweet spot.