Digital EQ Fact & Myth.

[PoorGlory]

I mainly use the Digi EQ3 and Elemental Audio's Eqium. To me, they sound different. Maybe it's the disconnect between the GUI and what's actually happening, I don't know... I understand the concept of why, theoretically, digital EQs should sound the same, but my ears say "not so much". Could be placebo effect, though...

[LPK]

Originally Posted by rhythminmind

So i decided to take it a step further.
How close could i get to the waves SSL with ANALOG on.

As I wrote in another thread:

The Analog Button on the Waves plugins does exactly one thing:

It adds noise, and nothing else (at least for API if I remember correctly, and definetly for the SSL EQ)
I've done the nulltest, and yes, it's just that.
No saturation, no this, no that, no nothing .. just noise.

Originally Posted by rhythminmind

The EQ variables that i am aware of are
frequency response
filter slope / Q
& phase

...

Besides linear phase, internal bit resolution, or to free cpu with external DSP i'm now questioning why anyone needs anything but there native EQ.
I can't find technical reasons.
Please inform.

The only reason, one would use an addional EQ-plugin is for "the feeling".
And this feeling comes from the GUI and the curves. This can bring you faster the results you might want, but it's not like such an EQ would provide a "higher quality" than a standart-parametric-EQ with oversampling.
And to spend like 300€ for an EQ like the Sonnox (or even 400€ for the Flux Epure) is just totally wasted money, because such an EQ is either already included in your sequencer or you can get a cheaper/free one (for example the one from ddmf).

The phase response of all EQs (as long as minimum phase) is the directly related to the curve. That means if EQ A provides the same curve as EQ B, the phase response is the same too.

[Dean Roddey]

I would have to be pretty badly hallucinating if the analog button on the SSL plugs don't also push up the lows a bit. If you have an SSL strip on each drum track and then have one on the master bus, and they all have the analog button on, the low end can be cranked up a lot, or so it sounds to me.

[LPK]

As you say it, I'm not sure not, if I have also tested the channel strip ... maybe I only tried it with the compressor and EQ of the bundle.

But it's really easy to test:

Load an audiofile, send it to two busses. Put on each bus one SSL plugin with exactly the same settings and swap the polarity for one of the busses.
The result should now be "no signal".

Now enable the Analog Button on one of the both EQs. Then you'll only hear the exact difference between using the plugin with/without the Anlog Mode enabled.

[CLH]

Originally Posted by Dean Roddey

I would have to be pretty badly hallucinating if the analog button on the SSL plugs don't also push up the lows a bit. If you have an SSL strip on each drum track and then have one on the master bus, and they all have the analog button on, the low end can be cranked up a lot, or so it sounds to me.

I'm sure that's right, and it would probably have to do with the noise or distortion added to the sound signal, that makes your ears perceive it as having more lows.

As to the essence of this thread, I think it's very interesting, although I'm not sure what to make of it or use it for. At the end of the day when producing or mixing, we all reach out for what works. And if plugging in X quickly gives me what I want, then why spend the time tweaking and comparing in order to make Y sound like X? Also, I'd have to X at hand in order for me to make Y sound like it...

What do you think?

[synthoid]

Originally Posted by LPK

The only reason, one would use an addional EQ-plugin is for "the feeling".

Well, no. One EQ might offer too few bands, too narrow a range of possibilities for setting bands and EQ, metering that is difficult to read, etc. Another might go above and beyond the call of ordinary EQs by offering extra filter modes, lots of configurable bands, interesting modulation possibilities, excellent metering, good presets, etc.

In digital audio, it's often that there is no comparative advantage between products at the level of sheer algorithm and mathematics. Think of convolution reverb for example. There's no doubt in my mind that the convolution algorithms underneath Space Designer and Altiverb will give bitwise identical output for identical inputs and impulse responses. Convolution "has a right answer", after all. But the impulse responses are not the same, and at the end of the day you come to realize that convolution reverb is all about the impulse responses; there's no comparative advantage to be obtained from the mathematics of convolution. And then there's the interface, user-configurable filters in the output sections, etc.

-synthoid

[LPK]

Ok, feeling and handling/possibilities.
But normally the EQ of your sequencer should offer enough.

[JP11]

I was always a bit bewildered by the excitement over the various digital EQs.

But inspired by this thread, I was comparing various EQs...some I could get to null..some I couldn't...it's my feeling that it's what you do with the EQ more than the EQ itself anyway.

But one thing I found odd, maybe someone can confirm this...in Cubase SX, the "lo" EQ does not null with itself...broke?

[chrisj]

Like I said on another forum's version of this thread, I have an EQ plugin coming out (I finally got serious about doing an Airwindows style channel EQ).

I am happy to say I improperly designed it in a way that to me sounds better than boring digital EQs that sound different variations of the same. I'd still use those for certain purposes, like if I had surgical-type corrections to make that weren't supposed to be interesting-sounding.

I've got a high cutoff that gets softer and woolier (nice on drums, not for everything though), two sweepable boosts that sound delicious (ought to, they behave similarly to analog tube EQ peak filters which I studied, and have to be synthesized via convolution), and a low cutoff that gets more punchy and kickin' as it takes effect (on the fringe of breaking up into sort of sputtering nonlinearities)

None of this is just saturation stages. It's all weirder than that. I'm happy with it and it should come out next week, for $60. AU, universal binary. (there's also a major surprise in store)

Feel free to skip a $500 glitzy biquad EQ filter product that's like the other biquad EQ filter products sonically to grab mine- I would not dream of trying to duplicate what is already very workable and ships with most DAWs, stock. I just don't think that stuff sounds good compared to what you can do with convolution kernels and coding up strange algorithms in the dead of night. If I grab an EQ for something, it's going to be mine and not the correctly-made Logic channel EQ, because I am never looking for a clean abstract correct changing of tonal balance- I want the sound to have more attitude if I'm throwing EQ on it.

[matt thomas]

Originally Posted by chrisj

(there's also a major surprise in store)
.

graphics?

narco

[JohnRoberts]

I guess this isn't widely discussed here but there is not a standard definition for how to define Q of boost/cut EQ sections. For typical HP/LP/BP sections, bandwidth is simply defined by the half-power or -3dB points, but in boost/cut sections there is no reliable half-power point since the nominal bandpass is added to or subtracted from the wideband allpass using different topologies, or algorithms in digital code.

This problem has existed forever with audio EQs but in analog signal processing, few trusted the repeatability or accuracy enough to notice. A dirty little secret about 1/3rd and 2/3rd octave graphic EQs is that those 1/3 and 2/3 octave sections don't all even nominally have the same transfer function.

This has become an issue in recent years due to customers trying to to use manufacturer's presets for speaker crossovers, that when they involved driver EQ weren't the same between different DSP platforms.

I submitted this to the AES standards committee last year. Their speaker group agreed it was a problem but kicked it back to the signal processing group to propose a solution. I hope they meet this week in SF so we could get a progress report. This may take a while, as it will turn out a little like the pin 2, vs. pin 3 hot standard that made half the industry wrong.

When comparing boost/cut EQ settings between different DSP platforms the amount of boost/cut, and center frequency of the boost/cut will agree precisely. Look for a difference in how wide or how narrow the region of boost/cut is.

I hope this clarifies at least one factor for sonic differences between EQs.

JR

[Alxi]

Will someone take the test a notch up and try to null UAD eqs with an other EQ ???

That would be interesting

-Alxi-

[Coyoteous]

There are also curve differences:

http://www.eiosis.com/images/amlt/EQCurveComparison.gif

[Kenton]

Decrampled is such a great word - I wish I knew what it meant.........

[matt thomas]

Originally Posted by Kenton

Decrampled is such a great word - I wish I knew what it meant.........

yeah, it gives 3 hits in google, and nothing in dictionary.com..

narco

[JohnRoberts]

I am not a DSP wonk so keep searching but IIRC the term was invented to describe a difference between LP analog filters and digital implementations due to inherent errors related to working with discrete samples and digital algorithms.

The significance of this surely varies with the application and magnitude of any variation.

I would keep searching for a real explanation from the company(s) pimping it as important.

JR

[Arksun]

Originally Posted by Kenton

Decrampled is such a great word - I wish I knew what it meant.........

If you go to the Eiosis site you'll see written that the decrampled line refers to eq's using the Orfanidis style filter (which many eq's use)

Whatever AirEQ uses its something different and seems to backup my experience with it. I love the sound of the AirEQ.

[zmix]

Originally Posted by j2j

I think I have a very simple argument here, proving the op incorrect.


Robin from www.rs-met.com wrote:
vieris wrote:

All properly designed Digital plug-in EQ's are the same.


which is a well known fact:

Document Not Found 0.1.1.23.3931

...assuming, we stick to the standard biquad implementation and demand the gain at Nyquist frequency to be 0 dB.




I don't know anything about c++, or coding computer software, but what are you saying Mr. Robin? There is a fundamental algorithm that parametric eq's share? That is what I think you are saying, that is what I think the OP is saying.



Let us chew on this concept--

Are all airplanes, cars, trains, lasers, surgical instruments, buildings, bridges and etc, the same? All of them share some fundamental concepts. Actually, everything on earth seems to share concepts.

I'm sorry, but to build a bridge you follow some very stringent mathematical formulas. Perhaps the same is for parametric eq's. I fail to see how all bridges are the same.

That said, " All eq's are exactly the same, except for X." Is a pretty weak argument. IMHO...


Even if it is slight difference in the distortion and q....


Well, What about the slight differences between a bullet, and a missile?

In a sense, it's like showing your groceries that you bought while driving a Ferrari, a beat up pickup truck, etc....

[zmix]

Originally Posted by rhythminmind

So i decided to take it a step further.
How close could i get to the waves SSL with ANALOG on....

Once and for all... the Waves "Analog" mode ONLY adds hiss at -106dBfs....

It does nothing else.

Period.

[zmix]

Originally Posted by JohnRoberts

I am not a DSP wonk so keep searching but IIRC the term was invented to describe a difference between LP analog filters and digital implementations due to inherent errors related to working with discrete samples and digital algorithms.

The significance of this surely varies with the application and magnitude of any variation.

I would keep searching for a real explanation from the company(s) pimping it as important.

JR

"Decramped" means that the filter will add poles and zeros (adding steepness) to the upper edge of the filter as the frequency approaches the nyquist limit.

This is very important to prevent aliasing as the frequency of the filter is increased.

That website which touts it's EQ as having "Analog Curves" must use either oversampling or decramping to avoid aliasing... however, it does not state that it uses either... I would expect that aliasing would be a problem here.

[rhythminmind]

Originally Posted by zmix

Once and for all... the Waves "Analog" mode ONLY adds hiss at -106dBfs....

It does nothing else.

Period.

Ahh ok this would explain the slight phase difference. Damn i was probably even closer matched with the Live EQ then i thought.

[zmix]

Originally Posted by rhythminmind

Ahh ok this would explain the slight phase difference. Damn i was probably even closer matched with the Live EQ then i thought.

Try it with the noise off..!

[matt thomas]

on their comparison chart they use the word "decrampled" whereas elsewhere they use "decramped". This will probably explain why there were no google results for "decrampled"

narco

[space2012]

Originally Posted by rhythminmind

...I have been able to recreate the the slope, phase, & sonic characteristic of the Duende & CSP with built in native EQ . Or with any fully variable parametric EQ for that matter. ... Besides linear phase, internal bit resolution, or to free cpu with external DSP i'm now questioning why anyone needs anything but there native EQ.
I can't find technical reasons.
Please inform.

more .mp3 320kbps and .wmaL
more screenshots

i have done similar tests, most plugins are in same level,

do a simple test,
import a .wav song from a cd, that is super recorded/mastered

duplicate that song/track
use native eq in A, another "SSL" in B,
do your magic,
then change the phase in one track,
if what you say its true, they should cancel 100% and nothing will come out in the master channel.
absolute silence.

if theres not 100% cancelation, its
probalby your Wordclock fault, or your DAC, or your cables, or your amplifier or your loudspeakers, or your headphone amp or your headphones!
or your ears or your testing methodology or the plugins.

plugins that have diferent algorithms, its verry unlikely that two diferent algorithms make 100% the same results.
waves req hishelf has totally diferent sound than kjaerhusaudio geq7 hi-shelf, and totally diferent than behringer ddx3216, and totally diferent than waves qclone, and totally diferent than roland vs2480, and totally diferent than free parisEQ1.1 vst, and totally diferent than izotope ozone, and totally diferent than sonictimeworks eq1, totally diferent than yamaha type1 and diferent than type2, etc...

or use a file comparator. they should be 100% equal, bit by bit.

[David Rick]

One reason most (non-linear phase) digital EQ's sound similar is that audio engineers expect the user interface to have the same cannonical controls: Gain, Freq, and Q. This leads designers down the path of bi-quadratic filters, which drastically reduces design freedom.

Having decided on cascaded bi-quads, the majority of programmers will pull down the same DSP cookbooks: Zölzer, Bristow-Johnson, and (if they want to do "decramping") Orfanidis.

Now back when everyone coded in fixed-point, there was still room for variation due to internal architecture. Different signal flow graphs produce different internal roundoff errors with these various errors being "noise-shaped" in different ways. But in practice, the optimal state space forms were seldom used, because they were less efficient than the standard formulations.

Now that many plug ins are being coded in floating point, internal architecture doesn't matter quite as much. But of course the programmer can intentionally put saturation features at certain internal nodes. The result will depend on the topology, but since these topologies may have little to do with analog topologies, it's not clear how this will satisfy the expectations of a typical user.

Digital EQ's don't have to have either minimum phase or linear phase -- other choices are possible. The question is whether these other choices are "musically useful." The truth is, our music has been shaped by minimum phase EQ's for so long that we might not even want something else.

If we restrict ourselves to minimum phase, interesting things are still possible. But to get them, we have to break free of the "tyranny of the bell curve". Think of how a graphic equalizer curve looks if you just push up one fader. Now imagine that curve having a different shape: it might have steeper sides, a broader top -- it might even be asymmetrical. There could be a knob for the plateau width, several choices for the skirt slope, and a knob to control the skew. It's really no problem to build such things in DSP, but the question is: would you use them?

Would you be willing to learn that part of your craft all over again? Or do like the fact that you already know how to use an EQ?

David L. Rick

[Football]

Originally Posted by chrisj

Like I said on another forum's version of this thread, I have an EQ plugin coming out (I finally got serious about doing an Airwindows style channel EQ).

I am happy to say I improperly designed it in a way that to me sounds better than boring digital EQs that sound different variations of the same. I'd still use those for certain purposes, like if I had surgical-type corrections to make that weren't supposed to be interesting-sounding.

I've got a high cutoff that gets softer and woolier (nice on drums, not for everything though), two sweepable boosts that sound delicious (ought to, they behave similarly to analog tube EQ peak filters which I studied, and have to be synthesized via convolution), and a low cutoff that gets more punchy and kickin' as it takes effect (on the fringe of breaking up into sort of sputtering nonlinearities)

None of this is just saturation stages. It's all weirder than that. I'm happy with it and it should come out next week, for $60. AU, universal binary. (there's also a major surprise in store)

Feel free to skip a $500 glitzy biquad EQ filter product that's like the other biquad EQ filter products sonically to grab mine- I would not dream of trying to duplicate what is already very workable and ships with most DAWs, stock. I just don't think that stuff sounds good compared to what you can do with convolution kernels and coding up strange algorithms in the dead of night. If I grab an EQ for something, it's going to be mine and not the correctly-made Logic channel EQ, because I am never looking for a clean abstract correct changing of tonal balance- I want the sound to have more attitude if I'm throwing EQ on it.

Sounds interesting. Look forward to checking it out.

[David Rick]

Gee, I guess I put everyone to sleep with nerdy DSP stuff before I got to this part:

Originally Posted by David Rick

If we restrict ourselves to minimum phase, interesting things are still possible. But to get them, we have to break free of the "tyranny of the bell curve". Think of how a graphic equalizer curve looks if you just push up one fader. Now imagine that curve having a different shape: it might have steeper sides, a broader top -- it might even be asymmetrical. There could be a knob for the plateau width, several choices for the skirt slope, and a knob to control the skew. It's really no problem to build such things in DSP, but the question is: would you use them?

Would you be willing to learn that part of your craft all over again? Or do like the fact that you already know how to use an EQ?

And to take the discussion in another direction, it seems like people value certain old hardware EQ's for sonic characteristics (distortion) some of which happen even with no boost or cut. Well, suppose you could "boost" a particular frequency region, but the plug in didn't really boost it -- it just created more 2nd or 3rd harmonics (but only for stuff in the "boosted" frequency region). Would that be useful. Does somebody already make such a plug in?

David L. Rick
Seventh String Recording

[zmix]

Originally Posted by David Rick

Gee, I guess I put everyone to sleep with nerdy DSP stuff before I got to this part:



And to take the discussion in another direction, it seems like people value certain old hardware EQ's for sonic characteristics (distortion) some of which happen even with no boost or cut. Well, suppose you could "boost" a particular frequency region, but the plug in didn't really boost it -- it just created more 2nd or 3rd harmonics (but only for stuff in the "boosted" frequency region). Would that be useful. Does somebody already make such a plug in?

David L. Rick
Seventh String Recording

David,
Au contrair! I found your description of the state of the EQ in the digital age very interesting, and well worded, too! In hardware, as in software so much of what informs the user's decision to choose one piece over another is simple convenience and /or familiarity...

As for your idea, it is hard to imagine the sound of a process that would only generate harmonics from signals within a specific frequency range without adding gain in that range.

[Ethan Winer]

Originally Posted by David Rick

Gee, I guess I put everyone to sleep with nerdy DSP stuff

Not at all! More please!

--Ethan

[Luny Tune]

Absolutely, David. Don't hold back!!

Regarding your "harmonics EQ"... A sort of "parametric exciter"...? Maybe it could work fine on a recording in poor shape and even add emphasis to a recorded signal already in good shape. For certain applications, in short. It sounds to me like an "eq" for saving signals rather than for making what's already good better.

Dunno. I guess we'd have to try it out to really know.

But working with good eq's already I'm not really missing anything, I must say. I believe that the limits for me lies within my talent and abilities rather than in what features and fuctionalities are available to me today.

I work with hardware eq's..., well hardware everything really, when I work with music. I use software eq's when working with voice over work simply because I'm able to save presets and all that. With ProTools' version III eq, Logic's version 8 eq, Cubase's version 4 eq (I've only testd this for a few days now) I feel absolutely no need to buy third party eq plugins. The ones I've tried don't outperform the three mentioned here in my opinion. I'm not at all surprised by the findings in this thread.