How do you judge an EQ?

[Zaphod (giancarlo)]

Honestly, I tried to build an example, but I stopped quite immediatly. It wasn't so obvious as expected.
Showing that a particular eq is not right (for bugs, approximations and other) is a bit off-topic and doesn't explain well how to judge an equalizer imho. Here the reason I stopped.

I agree with the sentence expressed 'there exstis 3 types... good, bad and regular' and with bmanic who said 'with a good eq I find quickly a result', it is pragmatic, but near to an answer, and with eliosound 'you should put it on your music for judging, otherwise they seems all similar'. In my opinion this topic infact is a bit too technical for a section dedicated to users. I was expecting examples: user wich say I tried this but it took 30 minutes for having this, instead using this I opened and it worked. Just to understand what happends in real life, but from the user point of view (expected in the user section)

As user (I'm an user too), I don't care diagrams, specifications, manuals. I put an eq on and I say 'I works' 'I should tweak' (after 10 minutes this sentence becomes 'I change') and 'It sucks'. I don't know why, but sintefex equalizers suffer the lack of precision, but (probably being a sample of the best analog ones, imperfect but born to be musical) give me faster results.

@christian: all.zip appears as corrupted, i can't open it.

back to work again bye bye

[Barbarossa]

but the problem with that is, that dsp engineers are better with math than with their ears. you need guys with a lot of experience to select and fine-tune settings.

indeed. even in this thread there have been comments proving this, ie. "it fights agaist my knowledge of signal theory". "phase shift causes distortion=bad"...

so what if it sounds good. and hey, they spent 40 years trying to remove distortion and hiss from tapes. Now people are dying to get it all back.

Totally wrong. YOU claimed, that higher phase shifts would preserve transients better, where i said, that this would be in oposition to signal theory and i asked for a theoretical explanation of your statement.

And you are also totally wrong if you think, that theory is a handicap in understanding things. It is a necessary tool and the most important tool for engineers. Without a working theory, no computer, no audio plugins and no EQ-plugins. It's all based on theories and they become more complex, the more detailed real things should be modelled.
Not less theory brings sucess, but a better understanding of reality. And that is reached by a working theory.

What you are advocating for, would lead to hi-end business, but not technical engineering.

I recommend you to read the "xhip is amazing" thread. There you have the example, how an excellent understanding of the technological aspects can lead to a good model and produce an absolute state of the art sound.

So it's not a matter of engineering/science OR musical hearing, but solid engineering AND excellent musical hearing giving the feedback for the adaptation of the theory.

[Zaphod (giancarlo)]

totally agreed barbarossa. Since math exists we have daws and plugins today. We are not adding numbers in a randomic order, we are following previous discovers and formulas

Just my point of view:
Sometimes theory is a sort of barrier for the discussion, and since it doesn't cover every aspect (we don't know EVERYTHING, for example we lack hyperspace technology) a discussion is always a good thing; sometimes you discover something new forgetting for a while what is 'wrong' and what is 'right' and listening new points of view. Maybe wrong point of view. And I'm not defending here kingston, but I agree with 'theory can't be the argument for every discussion'

I LOVE musicians that simply say 'wrong, right, I don't know, you said this is imperfect but believe me now it is working' more than developers that say 'this is the plot. Theory says it's bad. Believe me, you can't go far with this',
I'm comparing orange with apples, sorry

[bmanic]

The ABX software that I use can be found here:

http://www.kikeg.arrakis.es/winabx/

Cheers!
bManic

[xander]

Fatastic info in this thread -- even for a dope ike me!

[martian]

right, i assumed no-one tried my test because the file was too big or you dont like the download site? anyway ive done a better smaller version. hopefully this is a digital min phase/biquad/whatever apples apples comparison for 4 eqs in terms of freq response at least.

test curve:


zip file (8mb): http://s1.upload.sc/request/f10a6284c72 ... 2565c0cf0d
(sorry about the anoying site, choose to download then click 'here', done)

the interesting thing is that there are clear differences between the 2 that upsample to get the symetrical high freq bell (gliss and pohifopit) and the other 2 (waves ren and sx Q). as said before the 2 that upsample sound harsh and cold! more distortion from the oversampling? christian?

waves and q sound equal to me.

[martian]

The ABX software that I use can be found here:

http://www.kikeg.arrakis.es/winabx/

Cheers!
bManic

for some reason i used to have a problem running this, may try again now i have new system.

but i find a quick alternative is to just load up the 2 files in sx, mute one track then solo it (so as when you unsolo it goes muted again), keep the mouse pointer hovering over the solo button of that track and close your eyes or look down at keyboard (just keep ears positioned nicely), then randomly click a few times so you have no idea which is playing. identify each correctly 10/10.

[defjamm]

right, i assumed no-one tried my test because the file was too big or you dont like the download site? anyway ive done a better smaller version. hopefully this is a digital min phase/biquad/whatever apples apples comparison for 4 eqs in terms of freq response at least.

test curve:


zip file (8mb): http://s1.upload.sc/request/f10a6284c72 ... 2565c0cf0d
(sorry about the anoying site, choose to download then click 'here', done)

the interesting thing is that there are clear differences between the 2 that upsample to get the symetrical high freq bell (gliss and pohifopit) and the other 2 (waves ren and sx Q). as said before the 2 that upsample sound harsh and cold! more distortion from the oversampling? christian?

waves and q sound equal to me.

that's a good question about the upsampling...christian if there are better ways to implement it as far as quality do it please as an option(don't know if it would matter).
i believe ren and q use both orfanidis.

[bmanic]

The ABX software that I use can be found here:

http://www.kikeg.arrakis.es/winabx/

Cheers!
bManic

for some reason i used to have a problem running this, may try again now i have new system.

but i find a quick alternative is to just load up the 2 files in sx, mute one track then solo it (so as when you unsolo it goes muted again), keep the mouse pointer hovering over the solo button of that track and close your eyes or look down at keyboard (just keep ears positioned nicely), then randomly click a few times so you have no idea which is playing. identify each correctly 10/10.

I prefer a true ABX tester and I always do at least 30 tries. If I want 'real' results I do it 100 times (no, I'm not kidding!).

Cheers!
bManic

[Kingston]

Totally wrong. YOU claimed, that higher phase shifts would preserve transients better, where i said, that this would be in oposition to signal theory and i asked for a theoretical explanation of your statement.

Well there isn't any, only some listening test experiences (personal and various EQ threads around the net and magasines), but it has proven to work in all the cases. HarmoniEQ is a great example of this. I specifically specified the phase vs. transient implementation to Aleksey and he implemented it as one of the modes. I knew it would work, but unfortunately had nothing to back it up but the sound. Luckily not all developers are conservative.

As I said, it's possibly a psychoacoustic phenomenon and you're free to research it further. For me the fact that it works is proof enough.

Imagine the wealth of tricks that the oldschool designers know. This is just an isolated example, but the way I see it, DSP algorithms can be thought very similarly to creative electronic design "as black art". I mean, imagine a guitar amp that is a theoretically perfect implementation. It would sound incredibly boring. Often you hear these designers say things like "in light of theory it would seem ok, but it rarely works in practise".

I see no difference to DSP in that respect, having done a fair share of fairly advanced audio DSP myself. Really only the very basic blocks and their use is carved in stone. Step beyond that, and you're on your own.

[Christian Budde]

that's a good question about the upsampling...christian if there are better ways to implement it as far as quality do it please as an option(don't know if it would matter).
i believe ren and q use both orfanidis.

The oversampling in posihfopit can be turned off ("economy mode"). In fact the normal mode is only normal for the "analog" algorithm, because soft-distortion caused by the circuit is emulated. That distortion creates aliasing, if you don't oversample. For "digital" it is recomended to switch off the oversampler ("economy"). So you have a switch!
The oversampler itself is a polyphase implementation. This should result in minimum phase, but I have to admit, that I've never checked it so far.
Right now I'm over the weekend in hamburg with my girlfriend and I'm learning all over the time, so don'T expect to much answers.

Christian

[defjamm]

that's a good question about the upsampling...christian if there are better ways to implement it as far as quality do it please as an option(don't know if it would matter).
i believe ren and q use both orfanidis.

The oversampling in posihfopit can be turned off ("economy mode"). In fact the normal mode is only normal for the "analog" algorithm, because soft-distortion caused by the circuit is emulated. That distortion creates aliasing, if you don't oversample. For "digital" it is recomended to switch off the oversampler ("economy"). So you have a switch!
The oversampler itself is a polyphase implementation.

that's what i meant, i know what the controls do. i don't know much about dsp, just a question if there could be better ways to implement it.

[Debutante]

Okay... so the general talking points in this fantastic thread have been quite refined. My question: is all this is once again subject to taste?

In my usage so far I've pretty much used either of 2 or 3 differnt EQ's. This thread has opened my eyes to some very fine points on the subject, but I honestly still cannot define in my own work some of the problems people have encountered. One such problem is harmonic distortion that some EQ's are responsible for. At this point I'm not even sure what that sounds like. Obviously to some of you it's easily discernable. So, is it the case that it's always unwanted?... is it a nasty artefact to have in a composition from a techincal vantage point? Can it be used artfully, or is it akin to phasing and other no-no's?

Also can someone recommend an EQ that exhibits lots of harmonic distortion, or be bothered to put up some quick examples? CHEERS!

[Kingston]

My question: is all this is once again subject to taste?

very much so. as you can see even the actualy design process is down to taste with different schools of thought.

but I honestly still cannot define in my own work some of the problems people have encountered. One such problem is harmonic distortion that some EQ's are responsible for. At this point I'm not even sure what that sounds like. Obviously to some of you it's easily discernable. So, is it the case that it's always unwanted?... is it a nasty artefact to have in a composition from a techincal vantage point? Can it be used artfully, or is it akin to phasing and other no-no's?

harmonic distortion is one of those things again, down to taste and school of thought. In analog design even some tube guys go to great lenghts with minimising this, and some "will let the color take care of itself" (as I heard one designer blurb out).

In digital design we haven't the advantage that some component would automatically generate it, so we have to add it separately. Again, there's a wide range of how and what you achieve this with. It's one those blanket concepts that cover such a wide range of sounds that very little should be generalised. In practise it "mushes up and smoothens" things. Or with bad design and failing equipment it might be downright nastiness. Heck, all those stomp box overdrives and distortions are practically nothing but harmonic distortion as well.

Also can someone recommend an EQ that exhibits lots of harmonic distortion, or be bothered to put up some quick examples? CHEERS!

At least Voxengo HarmoniEQ comes to mind, and C. Budde did mention some posihfopit modes do this. Then there's that new one from, uhh, (forgot the name) spectrasonics?

It could be generalised that it's not the EQs job add harmonic distortion, and hence the current crop does this rather subtly (IMO a good thing).

[defjamm]

One EQ I like very much for mixing is the very old, outdated looking, but light on the CPU freeware A0 parametric EQ. This one seems to be very transient-friendly - is there anything special about it?

i have a theory why people might like it and why you like it...it's similar with hydratone, other convolution eqs(with hardware irs) and urs(when the filters are engaged). and it's the same with A0: rather high harmonic distortion(compared to 'cleaner' eqs like kjaerhus golden, sonalksis and gliss). you should download christians vst plugin analyser and try better see for yourself.

with A0, when sat is turned off and there is no gain on the filters, harmonic distortion is rather low. but when you now add gain or take gain away, there is an increase in harmonic distortion. sonalksis, kjaerhus and gliss don't have that(as much).

this was bullshit btw, i misinterpreted the graphs regarding harmonic distortion

@debutante: this time i'm right hehe. harmonie-eq and posihfopit/electri-q add harmonic distortion in some modes.