True Peak way above 0dB even with limiter applied?

[astralprojection]

good that you figured it out. it just HAD to be something in the fx chain.
an eq shouldnt cause 2-3db of extra volume though (which that screenshot looked like), it should just cause a peak or transient here and there; but not an overall volume boost of 2-3db.. that makes very little sense to me too. even if its in zero latency; the distortion from the HP filter shouldnt cause that either. screenshot of eq?

[noiseresearch]

Wait wait wait wait wait wait I just figured it out. I had an EQ after the limiter that was only set to remove frequencies below 20hz and above 18khz. I moved it before the limiter and am now getting the TP I have my limiter set to. That makes no sense though, how does an EQ that only removes frequencies cause the TP to go nuts as if there were no limiter? I am so confused yet so relieved that the issue is solved. What the heck!?

Limiter and similar tools should always the last piece in the chain. The EQ after is causing phase shifts and a redistribution of energy hence a different outcome.

[exist01]

Wait wait wait wait wait wait I just figured it out. I had an EQ after the limiter that was only set to remove frequencies below 20hz and above 18khz. I moved it before the limiter and am now getting the TP I have my limiter set to. That makes no sense though, how does an EQ that only removes frequencies cause the TP to go nuts as if there were no limiter? I am so confused yet so relieved that the issue is solved. What the heck!?

Limiter and similar tools should always the last piece in the chain. The EQ after is causing phase shifts and a redistribution of energy hence a different outcome.

Ooh I see. Well lesson learned!

[exist01]

good that you figured it out. it just HAD to be something in the fx chain.
an eq shouldnt cause 2-3db of extra volume though (which that screenshot looked like), it should just cause a peak or transient here and there; but not an overall volume boost of 2-3db.. that makes very little sense to me too. even if its in zero latency; the distortion from the HP filter shouldnt cause that either. screenshot of eq?

This is the EQ I had after the limiter - just cutting below 20hz and above 18khz:


Phase shift makes a lot of sense, because the EQ is doing effectively nothing to the audio. Also if I have the EQ do actually nothing, the limiter isn't affected. Only when I make tiny changes in the EQ does the limiter start misbehaving.

[noiseresearch]

good that you figured it out. it just HAD to be something in the fx chain.
an eq shouldnt cause 2-3db of extra volume though (which that screenshot looked like), it should just cause a peak or transient here and there; but not an overall volume boost of 2-3db.. that makes very little sense to me too. even if its in zero latency; the distortion from the HP filter shouldnt cause that either. screenshot of eq?

This is the EQ I had after the limiter - just cutting below 20hz and above 18khz:


Phase shift makes a lot of sense, because the EQ is doing effectively nothing to the audio. Also if I have the EQ do actually nothing, the limiter isn't affected. Only when I make tiny changes in the EQ does the limiter start misbehaving.

Phase is still a neglected topic especially by beginners. While it's often not audible per se it can damage a signal a lot. Like in your case. That's also a reason why I generally suggest not to run every instrument through highpass filters like some people say unless you know exactly why and what for.

[imrae]

Those filters look very steep, so it's no surprise that it would create some large phase ripples. Things to take away here are:

- Brickwall limiter goes at the end
- Expect sharp EQ filters to change your peaks

[astralprojection]

try changing to linear phase. the phase shift in zero lat does cause distortion too, especially in the low end. and its clearly audible if you for example use a steep HP filter on a kick. you can hear the clear difference from zero latency to lin phase where the latter sounds much better. but regardless of phase it shouldnt cause 2-3db of extra gain.

[DJ Warmonger]

What others said - always do HP with linear phase, as it won't change the amplitude. Also introduced phase shift is irrelevant at low frequencies.

Bah, I already made a video about it with examples.

[Kr3eM]

nvm

[mystran]

try changing to linear phase. the phase shift in zero lat does cause distortion too, especially in the low end. and its clearly audible if you for example use a steep HP filter on a kick. you can hear the clear difference from zero latency to lin phase where the latter sounds much better. but regardless of phase it shouldnt cause 2-3db of extra gain.

Well, there's technically no such thing as "zero latency" (although I do realize this is what Pro-Q apparently calls it) but rather "minimum-phase" which effectively means that the actual "latency" varies by frequency and is the minimum that makes the response possible. Since the "latency" (really "group delay") is frequency dependent, often "fractional samples" even in regions where it is more or less constant and often close to zero in some parts of the spectrum, these don't report latency, but there is still typically some (and more so the lower the frequency on which the filter acts). The varying latency over different frequencies is why these can distort the time-domain waveform quite a bit.

Linear-phase though will not automatically mean that the peak levels won't change (even though the shape of the waveform will probably change much less). Any peak is rarely the result of every frequency present in the signal being in-phase with each other, so removing some frequencies can raise the peak levels even if there is no phase-shift (after compensating for fixed latency in a linear-phase filter). Pretty much any processing whatsoever can change the peak levels in either direction to a certain extent, including things like lossy compression.

The right thing to do if you want tight peak control is simply always put your final limiter as the very last thing and (preferably) not push it too hard (ie. basically if you're pushing your limiter so hard that you feel you need an EQ afterwards, then you should probably tame the peaks first, then EQ and then do the final limiting with much less action).

[MogwaiBoy]

good that you figured it out. it just HAD to be something in the fx chain.
an eq shouldnt cause 2-3db of extra volume though (which that screenshot looked like), it should just cause a peak or transient here and there; but not an overall volume boost of 2-3db.. that makes very little sense to me too. even if its in zero latency; the distortion from the HP filter shouldnt cause that either. screenshot of eq?

This is the EQ I had after the limiter - just cutting below 20hz and above 18khz:


Phase shift makes a lot of sense, because the EQ is doing effectively nothing to the audio. Also if I have the EQ do actually nothing, the limiter isn't affected. Only when I make tiny changes in the EQ does the limiter start misbehaving.

Ouch - "40hz - 18kHz (Mastering)" is that a preset you got from somewhere? Don't listen to the EDM and trap kids on youtube, these kinds of cuts are not necessary - it's an internet myth. Only do it if you need to. But if you're gonna do it, put it before the limiter.

What that Pro-Q graph doesn't show you:


Even though you're not making it look visually resonant, there is still resonance due to the steepness. The high pass (low cut) is particularly concerning as frequencies around that high pass filter will peak higher, and are likely to be the main cause of your new true peak levels.

You hear it on youtube all the time "there's a lot of rumble energy down there you don't need, so you are safe to just cut it" - that is not universally true, and pays no attention to the effects of steep filters, especially on phase.

It is a sure way to kill the bass of your track though...

[plexuss]

Generally from a fidelity perspective shelves for rolling off the extremes sound better with as shallow a slope as possible. One reaosn is less resonance as per MogwaiBoy's post.

[MogwaiBoy]

Generally from a fidelity perspective shelves for rolling off the extremes sound better with as shallow a slope as possible. One reaosn is less resonance as per MogwaiBoy's post.

When I see/hear someone say "you can't hear under 20Hz so I put a brickwall cut there" I turn and I run in the other direction

6db/oct gentle sloping off, as you say - sounds so much better. I wish CraveEQ had 3db/Oct pass filters.