The Truth About Digital EQ

[Ermz]

Stumbled upon this very interesting thread on GS. I think a lot of the lower budget folks will be happy with the general conclusions drawn there.

http://www.gearslutz.com/board/so-much-gear-so-little-time/330874-digital-eq-fact-myth.html

 

[-Noodles-]

Bob Katz covers the subject in his book, and essentially covers (in more detail) points that are raised in that topic.

Really interesting when you really push a free EQ and hear / see the grain appear on screen - and then try one which 1000£/$ and find that it does the same thing at the same points, and the same gains.

 

[joeymusicguy]

very interesting

 

[Skinny Viking]

Oh my god my head hurts!

 

[GarethSE]

Pretty cool
I do notice differences in responsiveness between different EQ's though.
Before I started working towards going legit, I swore by Wavearts Trackplug for EQ, and Nuendo's built in EQ "Q" really doesn't do anything for me. I use it but it just doesn't feel quite as responsive. I can't feel the changes in the sound happening. It feels sludgy, like it just ate a big fuck off chocolate cake and had to move a bit slower otherwise it would throw up all over its shoes.

But the lack of differences between the actual effect of the EQ doesn't surprise me. I'm getting better sounding mixes than I ever did before with supposedly lesser quality plugins. I don't feel like the EQ is holding me back from making the best mixes I can possibly make, it's just a bit less pleasant to work with is all.

 

[Ermz]

I'm still reading... it gets hardcore right around page 3. From what I can decipher, most well-coded minimum phase EQs should function identically. The main differences being how accurate the GUI feedback is, and how conducive that is to your own workflow. Since going for the simpler Waves SSL pack, my work improved markedly, and I think that says a lot about how much I prefer a more straight forward, less cluttered, less option-laden GUI.

Might continue the thread after getting some sleep... Hope you guys get some use out of it!

Here's a post that encapsulates the conclusions that have been drawn:

If you ask me...
1) All minimum phase filters (that's more than 90% of filters you see) with the same frequency curve without added distortions sound the same. There is are no magical "EQ algorithms" that would make EQs sound different without added distortion or different frequency curves.
2) Some EQs are based on some specific hardware units. If they are, it's clearly stated somewhere, like those made by Waves, which seem to usually have 2-4 first harmonics. That means added distortion, nothing magical, very easy to implement.
3) Linear phase filters sound somewhat different from minimum phase. Apart from that, phase doesn't matter because all minimum phase filters have the same phase behaviour. You can add so called all-pass filters that change the phase, but this is inaudible unless you mix the original sound with the filtered.
4) Minimum phase filters can be made to sound the same. Some EQs have different filter shapes (URS N-series comes to mind) so you may need two or more filters of different shape to approximate it. For example, you could use one shelf and one bell filter to approximate a "pultec-style" shelf. The only difference is in the frequency domain.
5) When a minimum phase filter is said to be specifically "clean" or "accurate" it probably only has to do with the usual problem of that normally digital filters have +-0 difference at half of the sampling rate (like 22050hz) and that leads to less boost or cut at e.g. >18khz.

So, if someone comes to me and says that he has a new EQ design, I would ask him which is it: new filter shapes or added distortion. Most likely it would be just a new GUI.

 

[wyllie]

ouch....my brain hurts haha

 

[Erkan]

Why am I not surprised? Maybe compressors are different but I always get this feeling of most digital things being pretty much identical... even if not, atleast they're VERY similiar.

 

[Metaltastic]

Well, this is the kinda thread where I just read all your comments/impressions and take your word for it, rather than trying to make sense of it myself! But the findings sound good to me!

 

[Erkan]

Well, in digital domain, it's all about the mathematics... and I just don't see why different EQs would sound different when they all aim to do one and the same thing, and to get to that result there usually is a "correct" mathematic formula so they all end up using it. If an EQ sounds different, it's just because they added some other shit in it that you can just add after the EQ yourself, like saturation or whatever. Analog stuff has a gazillion more variables than a digital EQ...

I don't know, maybe it's a stupid example but it's like "Oh hey I'm gonna get the number 2 by adding two numbers together... now how should I do it? The simplest way to go would be adding 1 with 1 but I'm gonna be different so I'll take 100, divide it by 2, add 150 to that, then divide it all by 100. Oh crap, I got the SAME result as the first way to do it...". (I can has sense in my post? ¯\()/¯)

 

[NickL]

The problem is, not all digital EQs are designed to do the same exact thing. Many pieces of outboard analog equipment are revered not because they're super clean, but because they distort the sound in a way that is pleasant ("coloration.") So it's not unthinkable that tons of programmers are trying to work their owns versions of coloration into these programs, if not modeling vintage gear entirely. I really don't know where and how they go about getting the algorithms for digital equalizers, but I would assume that there are many more than just one standard equation floating around.


On a side note, I just discovered DDMF plugins in the thread that was posted. You choose what you pay for them, and they have a mostly positive reputation as being versatile. (Sort of the Radiohead of EQ plugins.) Does anyone know anything about these?

 

[Erkan]

The distortion that is applied in the EQ can be applied after the EQ as well, there is no need to bake it in the EQ. But that's what they do to sound different from eachother, but the EQ part is the same... if you read through that thread. So what you're paying money for is 2 plugins in 1, when you can just download either one of them separately for free or in more "pro" versions. When they smack 2 things together in one, they don't create something new... they just give the user something that already exists, but with less control over what's going on. This is probably appreciated by less experienced users who slam that EQ on a track and go "WOW, listen, it changed the sound completely without me tweaking it".... well guess why it does that.

 

[Ermz]

^ That's exactly it.

The Waves API and SSL series seem to slot into that category. It's hilarious that all the 'Analogue' button the SSL series EQ/Chan Strip does is add noise at -109dBFS. Even without the button engaged it still has the harmonic distortion.

For the 'cleaner' EQs, the main differences tend to be the curves. They can be roughly approximated in just about any minimum phase EQ, but obviously some will get you closer to curves you like right out of the box, with less work.

It's also great to see some of those crazy artifacts they measured, like the RenEQ with a filter that's too steep.

 

[NickL]

The "decrampled" conversation made me laugh

 

[Aaron Smith]

It's also great to see some of those crazy artifacts they measured, like the RenEQ with a filter that's too steep.

Wow, that is extremely interesting! I use the RenEQ all the time, and for quite a long time, I have noticed that pushing the filters too steep (some sources more than others, and more often during a hi-pass) can sound really weird- "phasey artifacts" is probably the best way I can describe it. I've never known what's actually been going on, and of course I haven't really needed to know since I'm relying on my ears anyway, but it's cool to finally crack open the issue a bit. I wonder why Waves hasn't fixed it? I can't really imagine an instance where anyone would actually want those artifacts there. It basically makes a sharp slope unusable in a lot of cases.

 

[slashvanyoung]

Wow, that is extremely interesting! I use the RenEQ all the time, and for quite a long time, I have noticed that pushing the filters too steep (some sources more than others, and more often during a hi-pass) can sound really weird- "phasey artifacts" is probably the best way I can describe it. I've never known what's actually been going on, and of course I haven't really needed to know since I'm relying on my ears anyway, but it's cool to finally crack open the issue a bit. I wonder why Waves hasn't fixed it? I can't really imagine an instance where anyone would actually want those artifacts there. It basically makes a sharp slope unusable in a lot of cases.

I've noticed this with several eq's so far. If you're pushing a steep slope/a certain band beyond a certain point it becomes noticeable. Not so cool if you want to push a certain band really hard...

Anyway, I gotta say I'd probably be much more surprised if there were major differences between digital eq's (soundwise), especially if they aren't modelled after a hardware unit.

Designing/coding a digital eq for "clean" frequency boosting/cutting is nowhere being rocket science... and that's what I expect from a digital eq: "clean" boosts/cuts and not so much colourization.

So ITE the GUI arguably plays a more important role. I can't stand working with the standard Q in Cubendo, but Reaper's ReaEQ is pretty neat IMO.

 

[Kazrog]

Great post. I've also read this in Bob Katz's book, but it's really a reminder to get things right outside the box before you start going crazy with EQ in the box.

 

[abyssofdreams]

Wow.

 

[Uladyne]

Well, this is the kinda thread where I just read all your comments/impressions and take your word for it, rather than trying to make sense of it myself! But the findings sound good to me!

+1

 

[Kazrog]

One great point raised near the end of the GS thread is that WORKFLOW is very important with EQs. I'm sure I could replicate my Neve or Pultec plugins with a good parametric EQ, but it would take a lot of work, whereas when I pull up a Neve or Pultec I get "that sound" right away and have more time to worry about the big picture.

If there was a flexible parametric EQ, however, that had tons of bands on tap and had presets such as "Neve," "Pultec," "API," etc. then it would be the ultimate. Anyone know of such a plugin (preferably for VST?)