So Mike are you saying that you think it"s phase cancellation "removal" that caused the peak out of no where?

Or is the phenomena you you talk about different?

And are you saying that a lack of understanding of this leads to a lot of "snake oil" solutions? 😊

Grem
So Mike are you saying that you think it"s phase cancellation "removal" that caused the peak out of no where?
 
Or is the phenomena you you talk about different?

 
I am agreeing with the explanation provided at the link in my OP as well as the similar explanations offered here. It took me while to remember that I'd heard this explanation before, and I don't think I ever recognized it happening to the extent that I observe it on this particular example, so I jumped to a supposition that there was an resonant hump and disregarded the possibility suggested by those who answered the question. I'm just emphasizing my guess that the effect may be easier to encounter with the relatively unique example of a crash cymbal and it's multiplicity of closely spaced frequencies.

Grem
And are you saying that a lack of understanding of this leads to a lot of "snake oil" solutions? 😊

Not quite. If I am saying anything, it is that an understanding of this can lead to many effective solutions, all of them relatively obvious. For example; "turn it down homer". :-)

I agree. I just meant to point out the phenomenon is common within the realm of EQing and K-level metering will not always be a safeguard against it. It's something worth to be aware of.

I think maybe some of it could be the way the plug-in is handling the input. Maybe a different design would behave differently. As Mike has said there seems to be an interrelated effect concerning key and harmonics which could be either hyped or lessened depending on that inter relatedness.
 
When  I get into a more dense mix or a mix with instruments that have unique harmonics there is more of a  need to watch how everything plays together. Mike mentioned drums...those are fast transients and they throw out a lot of spurious sound other than the original wave.Lots of harmonic after effects happening. I'm thinking mainly of the higher drums and cymbals. High and mid cut should get rid of a bunch of that but maybe not all of it.
 
Mixers sometimes intentionally add harmonics and induce the kinds of things we want to also remove at times. Sound is energy and the more energy you have happening the more you need to control it...comb filtering...phasing...all of those terms aside the energy clashes in some places more than others and in the ways defined. Those are just definitions for the effects.There's probably other things going on we can't entirely measure with our technology yet. 
 
If our ears are to sound as our eyes are to light we are missing most of the audible picture most of the time. 
 
It certainly looks like the levels rose a bit in Mikes example. How can we eliminate/understand it? In the case of that mix maybe just lower those ranges back to where you want them. A simple fix...not an explanation.
 
A simple explanation might be that harmonics re enforce each other and result in a cumulative boost. A really good EQ plug in might recognize harmful clashes and eliminate them or lower levels based on it, but it would need to know what you want and eliminate the stuff you don't.
 
 
 
 

Sanderxpander
I agree. I just meant to point out the phenomenon is common within the realm of EQing and K-level metering will not always be a safeguard against it. It's something worth to be aware of.

 
Sander is very wrong here.  Firstly this effect is not common to every EQ. I have many EQ's here which don't do it at all. 
( I even tried the crash cymbal source using Studio One's Pro EQ and still no change)  And if an EQ does create an upward level shift, K metering will always keep it under control simply by the fact the signal level is well under 0 dB FS and there is plenty of room for adjustment.
 
It's only a problem when your signals are very close to 0 dB FS eg peaks reaching say -1 dB  (and in Mike's case a 2 dB upward shift may cause clipping)  Although I doubt you would ever hear that either and as we know most DAW's have means to handle signals that go over.  Because K metering is so far away from that it does a great job of neutralising this problem and many many others too.
 
A much much bigger issue is that of correct gain staging within your DAW.  Something many many people just dont get. Simple as that.
 
I have always been aware of this issue but not encountered it and I think so for three reasons:
 
1  The EQ's I have been lucky enough to use don't do it or it is so little not worth worrying about.
2  As Mike interestingly has mentioned I have always used HPF EQ's on more musical tracks that don't obviously have the same harmonic structure as a crash cymbal.  (Mike, cymbals produce some fascinating harmonic structures as do toms as well) 
3  When I do apply a HPF to a crash cymbal I must be setting the cutoff frequency just high enough for it to not cause concern.  (30 Hz cutoff over a crash cymbal won't help you much!)  The secret to setting the cutoff of a HPF is where to stop.  There is a sweet spot where the low end of the sound does not change but sufficiently high enough to remove any unwanted subsonic rumble etc..Your ears will tell you where that is.

Jeff, it is my understanding that phase interference will cause this effect no matter which eq you use, simply due to the way the process works. If an EQ doesn't do it, it would be inhibiting frequencies other than the intended one. But I admit I'm not a technical expert on this, perhaps I'm missing something in the explanation given to me. Here's a SoS article on it;
http://www.soundonsound.c...articles/qa-1210-5.htm
Here's another apparent expert opinion that seems to back up what I'm saying
http://sound.stackexchang...ncreasing-output-level
As for K-level metering, again, I don't use it, but if the audio file is hot and the EQ creates a small level boost (2dB or even 1 or less) as far as I understand it the EQ could peak before you hit the meter. Or do you gain down audio files pre FX when using K-level metering?




Post culled by spam filter, now restored. K.

Ok somehow my post disappeared after editing it.

Jeff, with all due respect, I don't think I'm wrong at all about this effect being very common within EQing. Nor is it a sign of a bad EQ. Many expert articles seem to back me up on the technical side of this, I will admit I'm not an expert myself. Fundamentally, an EQ is changing a waveform. That may result in different peaks. The term "cutting" makes this counterintuitive but no less true.

As for K-level metering, again I don't use it, but it seems to me it happens at the meter/fader level, so any audio clips close to 0dBfs could still clip EQs in the insert path, no? Or does K-level stuff mean downgaining any audio files to add 14dBs of headroom?

EDIT:
Here's a link to an article with a nice graph and explanation.
http://sound.stackexchang...ncreasing-output-level

Hi Sander.  You are correct in what you say about tracks still being too loud at fader level etc and yes this problem will occur if that happens.  But as many people just don't get either, you track using the K system metering as well.  Although K system does not necessarily suggest you do that,  there is no reason why you cannot either.  This is where many go wrong with gain staging.  People are just not tracking correctly.  When you do, all your tracks are already at the correct K level before you even go any further down the line.  Hence many problems are just not there to start with.
 
When you track correctly it is painless to get a perfect mix at the correct ref level on all your buses too and hence your final mix.  It makes it all easier that is all.

That makes sense, so all that's left as a source for trouble is low quality/badly processed samples I suppose. Or super hot softsynths (Z3TA+ and Trillian come to mind).

EDIT: I'm correct about the other thing too by the way. Obviously I didn't say "every EQ" because I haven't tried every EQ. I said it's common within EQing and I would expect any accurate EQ without some kind of auto leveling to display this effect (depending on the sources and the specific EQ values).

Sanderxpander
...this effect being [is] very common within EQing. Nor is it a sign of a bad EQ. Many expert articles seem to back me up on the technical side of this, I will admit I'm not an expert myself. Fundamentally, an EQ is changing a waveform. That may result in different peaks. The term "cutting" makes this counterintuitive but no less true.

 
The Fabfilter has a 48dB/octave slope option, and when you use it you see that the material below the cutoff point is very effectively cut. When you use a typical EQ with a slope of 6dB/octave, or 12dB/octave you don't actually cut out very much energy below the cutoff point.
 
If you don't actually cut out very much energy, you don't actually change the components of the Fourier series very much and so the results of a change in the interference filtering may be minimized.
 
When you actually cut the energy, you actually change the components of the Fourier series enough to see the results of a change in the interference filtering.
 

Sanderxpander
As for K-level metering, again I don't use it, but it seems to me it happens at the meter/fader level, so any audio clips close to 0dBfs could still clip EQs in the insert path, no? Or does K-level stuff mean downgaining any audio files to add 14dBs of headroom?

 
K Metering was proposed as a system to calibrate listening levels. The idea is to select a target level on your meter, use the target consistently, and coordinate playback at that target with a Sound Pressure Level while listening to playback. The idea is that the consistent experience will allow you to preserve your ability to recognize dynamics as you mix and you will have lots of digital head room for transient peaks that rise above your average levels.
 
If you turn down your pre fx levels so that the fx outputs stay well below clipping, that is exactly what you are doing. The only relationship track levels have with K metering is that you have may a K-meter running on the master/main outs bus.
 
Anyone who reads the A.E.S. paper will see that K-metering is a "system" specifically describing the calibration of digital meters with SPL listening levels.