EQ for Phase Response


#1

I have an unconventional way of thinking about EQ.

Where most people would say that a given frequency is too loud, I identify that frequency as arriving to soon. Where most people would say that a given frequency is too quiet, I identify that frequency as arriving too late.

I make this assertion because equalizers cause phase shift. I am not talking about FIR filters, I am talking about IIR filters. If you doubt that humans can hear phase shift, please experiment with JUCE’s biquad all-pass filter module.

Does anyone else think about EQ this way?


#2

No, and it’s really really wrong. What does time have to do with the amplitude?


#3

You definitely can‘t say that. While a filter can alter both the phase and magnitude response and both effects are more or less noticeable, there are filters like FIR based linear phase EQs which alter the magnitude whithout delaying one single frequency more than another. If things would be as you said, how should that work?


#4

You can think that way. Only phase and amplitude together give a meaningful signal.

However, our hearing is by evolution not trained to perceive the phase of a signal. Only with a reference, there is information conveyed, e.g. in a stereo field, when the phase from left ear to right ear differs, it gives us hints about the origin of the sound.
And this is the reason, why in some mix stages linear phase EQs are important.

The second reason is that if your signal comes from different microphones, they pass different filters. So when mixed together, it creates artefacts. That is, why in an orchestra recording with main mics and aux mics per instrument, I would insist on linear phase EQs, vs in a close miked rock band, I usually don’t care and go for the less CPU expensive ones.

When I hear your description, it rather sounds like you want to shape the ADSR envelope of individual frequency bands, which could be an interesting effect.