Ah Xoc Kin wrote: realmarco wrote:
thanks for the link
You're welcome. Be sure to read Dave Gamble's comment in that same page.
That's me. I also built EQuick, EQuality, and way back when, did the 517Mk2 at Sonalksis. I've been designing EQs for a long time. I even designed the EQ in the Liquid Channel.
The "do all EQs sound the same" question is a good one, because the less you care about the answer, the more the answer is "YEP", and the more you care, the more the answer is "NO".
I care a lot about EQ.
It is definitely true that the range of spectral responses that can be derived from second-order digital stages (what shapes a digital EQ can make... without spending 'more than the usual' amount of CPU) is a fixed thing. But even then, how the EQ maps your controls into an EQ curve is immensely variable. Most digital EQs in history did this very badly. The ones mentioned in this thread are, by and large, the ones that got it more right than most. I think that article references the Orfanidis stuff, which (for bells) means that your top-end doesn't get completely ruined. The Orfanidis paper doesn't do a great job, and there are improvements which are in common usage that do better. I've been staring at this problem for more than a decade and I have techniques that go a long long way above and beyond whatever else is out there (as you can see by looking at response comparisons between EQuick/EQuality and anything else).
Here we're really talking about IIR-based EQs - the low CPU option. As soon as you are happy to spend the CPU on an FFT, there's just no reason why the response should be wacky at the top-end. The low-cpu case is the hard task.
So, there's that. What about if we're talking in a world where everyone gets the EQ mappings right?
Well, there's still interaction between the controls. This is the thing that you can largely ignore, to make the claim that all EQs /can/ sound the same. Trouble is, you can only ignore it if the EQs you're using have massive ranges (enough to copy whatever shape you needed). And actually, life is a lot faster with an EQ with good interactivity.
Mojo... what's that then... It's a combination of interaction between the band controls (for instance, there's a classic EQ whose shelf increases in resonance as you ramp up the gain), and interaction between bands (Pultec).
Actually, I've given up on thinking of the Pultec low shelf as two bands, because their interaction is SO extreme. They're sorta working in parallel, but not quite. It's one band with four controls. Can you model it with a single second-order low-shelf in EQuality? Yes, naturally, but it's much more fun poking at the knobs, not knowing what it's doing (and, frankly, it's so weird that it's very hard to predict).
There's Mojo from distortion, but I've STILL not found a schematic where the distortion is integral to the EQ section (though for synth filters, it's the norm), so following your EQ with a saturation stage is no loss of generality.
Anyway, I'm on thread. Any questions I can answer?