Digi EQ3 Plug + Level Increase w/ High Pass Filter

[UnlikeKurt]

The following has occurred when using the High Pass Filter included in the Digidesign Eq3 plugin.

When I select the High Pass as one of my bands, I immediately see that the output meter of the plugin shoots up past that of the input.
This is without any additional adjustments of that band, any other band, or the input/output levels.

This has happened with tracks ranging from bass guitar (where there'd be a good deal of actual material roll-off going on [for experimenting with this occurrence]) to something like glockenspiel with very little to no content down below

Now I was under the impression that, if anything, I would see a level decrease coming out of the plugin after cutting off a good chunk of low end power.

Has anyone else experienced this?
Anyone have any ideas to why this would occur?

Thanks all
James

[@?,*???&?]

It's not just the EQ3, it's any plugin.

I've noticed this with the MCDSP Filterbank when high-passing some hip hop tracks.

I suspect it's because you're removing alot of cumbersome artifacts below the HPF frequency.

Weird though, huh?

[UnlikeKurt]

Right, but I'd expect that the removal of "cumbersome artifacts" would result in a drop in output level, not a drastic increase!?

[hobbycore]

If there is compression of some sort down the chain then it makes sense to an extent.

[logancircle]

It's not just the EQ3, it's any plugin.

Yep. It doesn't happen once you've got the plug-in on the track or when/if you automate that plugin to bypass or whatever.

[UnlikeKurt]

Hobbycore: You've intrigued me, please expand on this. I'm not talking about the overall track level increasing. I'm speaking purely of the output level of the eq plugin itself. The input level of the plugin remains the same, and with high pass engaged, the plugin output level exceeds the input by a good deal. One would think the reverse would be happening as you roll off.

logancircle:
I'm a touch confused.
"it doesn't happen once you've got the plug-in on the track"?
Please explain what you mean by that.

I select the plug-in as one of the inserts on a track.
I then playback the audio and adjust the parameters of the plugin.
If I select the high pass filter as one of my bands, even without making any further adjustments (ie: boosts elsewhere), the output level of the plugin increases past that of the input at the plugin.

[digitaldrummer]

I've also noticed that with EQ3 - I had not pinpointed it as the HPF. I just got in the habit of backing the output level off a bit when I use it and notice "red". and btw, I don't usually see that with other plugins and I'm not recording super hot or anything.

[logancircle]

logancircle:
I'm a touch confused.

Sorry. All I meant was that once the plugin (insert) is activate and affecting the track everything is groovy. It's only when you first turn it on that it makes that weird sound. It's not always the same sound, either. Sometimes it's just the sound of a filter snapping on, which is annoying but livable. But other times is does a high-pitched whoosh. No idea why that is, which may be the whole issue here... ?

[@?,*???&?]

I've noticed this phenomenon specifically on a level maxmized track. The rappers bring in industry beats which are crushed. When you high-pass the sub frequencies, the track clips, badly.

Very strange indeed.

[logancircle]

...specifically on a level maxmized track...Very strange indeed.

I like that we all know what you mean by a level-maximized track. "We've maximized the level, sir. Our work's done here!"

[soundandform]

A regular digital meter doesn't show the true, underlying analog signal. You would need to use an "oversampling meter" to see the absolute peaks which can be hidden in between the actual sampling points. When you apply a high-pass filter you're also causing a phase shift. The shifting of higher frequencies will un-conceal some of these peaks, so they're now registered on the meter.

Steve
Massey Plugins Inc.

[soundandform]

On further thought I don't think my first answer is the core reason of this phenomenon, though it could partially help account for it. Instead, I think the phase-shifting of the filter simply reshapes the signal and the new signal can and will have different instantaneous peak amplitudes. Imagine the audio signal as higher frequency vibrations riding on top of larger lower-frequency waves. Now, by shifting the frequencies relative to each other, it's possible that the peak of a high frequency vibration correlates more with the peak of that lower frequency signal it's riding on. The end result is a composite waveform with a larger absolute amplitude.

STeve

[logancircle]

Hey Steve--thanks for Tape Head! And THC! And your limiter! They've saved a number of my mixes. Ethan Winer digs Tape Head, too.

[Professor]

I'm going to toss out a couple of possibilities for what you're seeing there.
First, is that when you engage a roll-off filter, it causes a slight bump just above the corner frequency. The steeper the roll-off, like as you kick from 6dB/octave to 12, to 18, then that little bump gets slightly more pronounced. It's not much, maybe 1-3dB or so, but that's enough, and it's usually spread across about an octave or so. That's a quirk of analog filters that has been built into the digital gear too either by choice or because it's just hard to avoid. So if you have a roll-off cornered at say 100Hz at -12dB/octave, you can expect a little boost from about 100-200Hz or so. I won't try to put an exaact number on it, but it's there. And of course if you have a track that is already "level maximized", then that extra bump in the bass or low-mids would be enough to put you into the red.
The other possibility I see would be that the meter is reading a root-mean-square (RMS) of the energy across the entire bandwidth from 20-20k or maybe even a little more. It's not giving a spectrum analysis, so all that data from all of those bands are being represented by one single meter. It seems odd that taking away a bunch of the sound energy would make the meter go up, but if that energy was dominating the average and because of phasing, cancellation, or some other weirdness, then removing that energy would allow the meter to respond more accurately to the material above the roll-off. Kind of like how you can sometimes hear distortion in the high-mids or highs when you turn up an EQ even though the meter reads below the red. The overall average of the signal is indeed below clipping, but in that particular band, the signal is going over, and we're able to hear it even though we can't see it. That's why it would be nice if every meter could instead be a spectrum analyzer... but then "level maximized' material would turn into 'spectrum maximized' material and all we would hear is white noise.

Those are my two best guesses at why we see that, and actually I'd kinda figure it to be a combination of the two.

-Jeremy

[mwingerski]

This is a reconstruction error. Pro Tools meters are inaccurate and don't catch these errors.

It is the main reason why you want to record digital signals well below 0dbfs

Read about it here...

http://www.gearslutz.com/board/masterin ... indle.html