Running an EQ into Plugin Doctor kind of shows how an EQ affects the ADSR, but it doesn’t show the full picture.
Already we can observe how a sine wave of varying amplitudes is affected, but to fully understand this we need a pure impulse file.
This is essentially white noise trimmed down to as short as possible. Using RX editor and tweaking each sample, I was able to make a short burst of sound that contains all frequencies with equal amplitude.
This is what it sounds like: play drac file
Notice it’s just a quick pop sound.
What do you think would happen if I equalized the signal?
For example, say I use aggressive filters on every C note. It should sound like the note C, but shouldn’t be longer than the original impulse - afterall I’m not changing the timing, just the frequency response.
But listen to what happens: Play before and after
When using zero latency EQ, the original impulse that should be more or less one sample, now has a completely different ADSR.
The sustain of the impulse makes it no longer an impulse.
Linear phase fixes this issue, but introduces another - commonly called pre-ringing, you’ll notice a swooping sound before and after the impulse. This becomes worse the longer the latency.
Watch the video to learn more >
So what’s happening?
With zero-latency EQ, the phase shifts needed to create the filters are great enough to significantly prolong the signal.
Since some aspects are unaffected and not shifted, yet others are, the processed signal now needs to extend from the start of the signal to the very end of the time required to create these aggressive filters.
As a result, the processed signal is significantly longer than it was before.
As for linear phase, the filter used shifts this post ringing to both pre and post ringing, causing a similar effect but before and after the original signal.
The amplitude of the ringing is also increased.
As you might have imagined, the lower the amplitude of these filters, the less significant the distortion.
For example, when I lower the filters to 10dB, the ringing decreases for both ZL and linear phase filters.
Lowered to 6dB, and the ringing is decreased further, and so on.
Lastly, attenuation doesn’t cause the same effect. For example, if I took all these filters and made them -30dB, we’ll notice the ringing is reduced significantly. However, it’s still present and even easy to hear.
Watch the video to learn more >
So, practically speaking, what’s the takeaway from this?
The more aggressive your filters, the more you’ll affect the ADSR of the processed signal due to phase shifts.
Additive filters exacerbate this issue, but subtractive filters still have this problem.
Hopefully, this makes you reconsider introducing aggressive filters at all - granted, I was showing an extreme example, but imagine the effect if you used 10dB filters on multiple instruments.
The effect adds up and may be so gradual as you mix that you don’t even notice it but can tell something is off by the end.
Soothe 2 is a useful plugin, but seeing how some people use it makes me cringe a bit.
I saw a video saying use this one trick on every mix - that is to side-chain the vocal while soothe 2 is on the beat and then cut aggressively.
That’s terrible advice - don’t do it. As we saw here, high q value high amplitude changes to the signal drastically distort it.
You could do it subtly, the negative effects will be mild that way, but the more aggressive and numerous these cuts, the more you’d completely change the ADSR of your signal by adding this ringing distortion.
It’s a quick fix that isn’t worth it.
Zero latency filters cause phase cancellation when used differently on similar signals.
I’m sure you’ve heard this before - if I add a HP filter to one track of a multitracked instrument, it’ll affect the phase relationship between them.
But I’m not sure if this concept has been shown.
In Plugin Doctor I can use the oscilloscope to observe what happens to the waveform as I introduce filters.
If centered perfectly on the affected frequency, the shift in the rotation is minimal, but as soon as I affect the frequencies below or above, notice how aggressively the signal is shifting in time. Again, the greater the amplitude change, the more aggressive the shift.
Since I doubt any of us are making music that’s one sine wave, this will always occur when we EQ - since frequencies above and below the center frequency will be shifted.
If I have a nearly identical signal that isn’t equalized, say, another mic on a guitar cab or something, this phase shift greatly impacts the phase relationship, resulting in phase cancellation.
Here’s what that sounds like.
Watch the video to learn more >
Don’t use aggressive EQ filters unless you’re okay with the negative effects it has on the ADSR, and the ringing and/or pre-ringing distortion it introduces.
Aggressive filters in similar processors will have a similar effect - Soothe 2 is a great plugin, but using multiple filters aggressively can screw up a mix quickly, which is something Soothe 2 does unless you’re careful with the settings.
Lastly, aggressive filters mean more phase cancellation between similar signals.
If you need to use different EQ filters on a multi-tracked instrument, linear phase is the best option, but as we’ve discussed, it also has drawbacks.