The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Non Linear FIR vs IIR Equalizer Plugins
Old 15th March 2016
  #31
Lives for gear
 

Old 15th March 2016
  #32
Audio Alchemist
 
Lagerfeldt's Avatar
Thanks for the pics!

So the DMG mode called "Natural" would be the equivalent of the "Natural Phase" in Pro-Q? I'm not sure if you matched the setttings, but the DMG one seems to have more ringing, but that might simply be a settings mismatch rather some inherent in the algo.

Quote:
Originally Posted by msmucr View Post
Digital+ and Pro-Q natural phase are different concepts..
As it was already mentioned and it is clear from Dave's quote, first one is compensation of IIR fiters, with Digital+ curve becomes closer to ideal magnitude respone of analog fiter, with Digital+Phase enabled, also phase response is affected by this.. but not to the same extent as in case of Equilibrium FIR filter mode with analog response.. there is global progressive phase shift towards Nyquist.

Whereas Pro-Q approach there is rather conventional oversampling with LP FIR fiters.. with inherent preringing and latency.

I'm not saying, what is better or worse, just to discern two different approaches, where one is really unique to me.

Michal
Makes sense. I'm not digging the Natural Phase mode in Pro-Q which could be the pre-ringing. But Digital+Phase seems to be different and at least on paper it seems superior.
Old 15th March 2016
  #33
Lives for gear
 
karumba's Avatar
i've just re-done the test i did some time ago with equality vs. pro-Q2.
it was the equality analog phase mode that matches pro-Q2 natural phase mode. the default Q (0.71) in equality corresponds to a Q of 1.0000 in pro-Q2. in that case phase difference test gives a residual below -120dBFS. so i would say the technical implementation seems to be comparable.

when i demoed equilibrium afaik i got the same result when using FIR mode + analog phase. here you also need to set the Qs differently to get the same response curve.

attached figure shows phase response difference near nyquist (blue=equality | red=proQ2).

//edit: added the time response. seems they are on top of each other.
Attached Thumbnails
Non Linear FIR vs IIR-equ_vs_proq2.png   Non Linear FIR vs IIR-equ_vs_proq2_time.png  

Last edited by karumba; 15th March 2016 at 01:54 PM..
Old 15th March 2016
  #34
Audio Alchemist
 
Lagerfeldt's Avatar
Is that the Digital+Phase mode?
Old 15th March 2016
  #35
Lives for gear
 
karumba's Avatar
the mode that is compareable to proQ2's natural phase is called "Analogue Phase" in equality.
(sorry, we where talking about equilibrium, but i meant equality.)

@msmucr i'm wondering why in your figures the impulse response of proQ2 natural phase vs. equilibrium analogue phase look differently. in my understanding equilibrium analogue phase is comparable implementation as equality analogue phase. did you take the different Q-definition into account?

Last edited by karumba; 15th March 2016 at 11:46 AM..
Old 15th March 2016
  #36
Lives for gear
 

Quote:
Originally Posted by Lagerfeldt View Post
Thanks for the pics!

So the DMG mode called "Natural" would be the equivalent of the "Natural Phase" in Pro-Q? I'm not sure if you matched the setttings, but the DMG one seems to have more ringing, but that might simply be a settings mismatch rather some inherent in the algo.
I'm not sure, if I can recall any "Natural" mode in Equilibrium..

Natural phase mode in Pro-Q is oversampling.. and pre-ringing is there due to sharp LP anti-aliasing filter in oversampler, you'll get very similar response from most of oversampled plugins out there.. like TDR SlickEQ or GML EQ in 44/48k modes.

Whereas Dave tried to avoid of oversampling usage in his EQs.. (as he stated somewhere.. sorry can't recall the exact quote). So he uses at least two different ways to get closer to the ideal and EQ always operates at host sample rate.

One is selectable FIR processing mode with Analogue phase.. its magnitude and phase response, gets closest to Pro-Q natural mode to me and also Jan verified at his previous post.
Although impulse responses looks different (1st vs 5th pic). Pre-ringing in case of Equlibrium analog is directly caused by its EQ FIR filter impulse response, whereas Pro-Q pre-ringing is there due to mentioned anti-aliasing filter, not directly because of used EQ filtering. That's why you see those differences in ringing, although EQ settings were very meticulously matched.

Second approach is use of his "corrected" IIR filters. Digital+ matches magnitude closer to the ideal (detail in 6th pic, yellow plot is with this feature enabled). Digital+Phase mode tries also to fix phase response, because normal IIR filter phase response curve is always warped towards zero value at Nyquist frequency (kinda like magnitude curves without any de-cramping).
But according to my measurement, when +Phase mode is on, it will also distorts magnitude response slightly and there is cca +0.3dB gain for whole audio band (not sure if this can play some role, when doing some subjective comparison between modes). It is possible to see that from my 9th pic. Additionally, whole phase response at this mode is shifted to -180deg towards Nyquist.. I cant remeasure it, because I've did all that with DMG trial previously, but -180 is so "nice" number , that it looks like uncompensated 1 spl plugin delay, when Equilibrium operates at this mode..
So pay attention to that, if it is used just on 2-bus, it won't be causing any problems, but for example, when it will be used at multi-miked or parallel tracks, all highs at the EQed one will progressively goes into antiphase with the rest.

Quote:
Originally Posted by Lagerfeldt View Post
Makes sense. I'm not digging the Natural Phase mode in Pro-Q which could be the pre-ringing. But Digital+Phase seems to be different and at least on paper it seems superior.
Well.. observed pre-ringing is always subjective and material dependent. In case of Equilibrium, you may also try zero latency FIR analogue mode, which doesn't have any pre-ringing.

Michal
Old 15th March 2016
  #37
Lives for gear
 
karumba's Avatar
Quote:
Originally Posted by msmucr View Post
One is selectable FIR processing mode with Analogue phase.. its magnitude and phase response, gets closest to Pro-Q natural mode to me and also Jan verified at his previous post.
Although impulse responses looks different (1st vs 5th pic). Pre-ringing in case of Equlibrium analog is directly caused by its EQ FIR filter impulse response, whereas Pro-Q pre-ringing is there due to mentioned anti-aliasing filter, not directly because of used EQ filtering. That's why you see those differences in ringing, although EQ settings were very meticulously matched.
thanks for confirming. of course, i wasn't assuming you didn't match the settings, i was only curious why your impulse responses for analog phase vs natural phase differ from my measurement results.

Quote:
Second approach is use of his "corrected" IIR filters. Digital+ matches magnitude closer to the ideal (detail in 6th pic, yellow plot is with this feature enabled). Digital+Phase mode tries also to fix phase response, because normal IIR filter phase response curve is always warped towards zero value at Nyquist frequency (kinda like magnitude curves without any de-cramping).
is my understanding correct, that digital+phase implements some kind of "IIR frequency+phase decramping"? in my understanding if the frequency domain is "IIR decramped", shouldn't the phase also be "decramped" automatically? in my understanding the desired result is to have an analog like frequency & phase response without pre-ringing, similar to something you would get when having an ideal analog EQ which is AD/DA'ed with an ideal converter.
Old 15th March 2016
  #38
Audio Alchemist
 
Lagerfeldt's Avatar
Thanks for the detailed clarifications, Michal - the fog is lifting! ;-) I wish this type of information was available in the manual since it's hard to know exactly what's going on in the different modes.
Old 15th March 2016
  #39
The pre-ringing you get from an oversampled EQ is less than the pre-ringing you'd get from even the best analogue loop (because one filter can be skipped and only the "change" really needs to be oversampled).

Nothing to be afraid of.


BTW Vlad once wrote a quick and dirty article about the subject here: https://vladgsound.wordpress.com/201...alizers-draft/

Last edited by FabienTDR; 15th March 2016 at 06:50 PM..
Old 15th March 2016
  #40
Lives for gear
 

Quote:
Originally Posted by karumba View Post
...
is my understanding correct, that digital+phase implements some kind of "IIR frequency+phase decramping"? in my understanding if the frequency domain is "IIR decramped", shouldn't the phase also be "decramped" automatically? in my understanding the desired result is to have an analog like frequency & phase response without pre-ringing, similar to something you would get when having an ideal analog EQ which is AD/DA'ed with an ideal converter.
Yes, this special mode try to address both magnitude and phase wrapping.
IIR filters un-warping of magnitude doesn't automatically mean, phase response is also affected. So far I came to several implementations of filters in plugins and digital hardware.
- IIR filters, where standard textbook bilin. transforms are used to calculate filter coefficients.. Both gain and phase shift are zero at nyquist. Those are most common.
- IIR filters with prescribed gain, so modified formulas for coeff. calculation with additional input parameter (gain at Nyquist for ideal filter) are used. Those are mainly based on paper by S. J. Orfandis. In this case gain at fs/2 equals to pre-calculated value, but phase response is still progressively warped towards zero. And magnitude response is still tiny bit different to analog ideal, you can see it also visually from magnitude plots, where resulting shape approaches fs/2 parallel to x-axis instead of ideal angle.
This is what is you'll get with most modern EQs.. like Pro-Q zero latency, Equilibrium IIR without any compensation enabled, Oxford EQ.
- DMG's exclusive IIR filter with additional FIR compensation Digital+ further corrects magnitude response and Digital+Phase also phase response with few small remarks, I mentioned before.

So if it is better or worse is probably bit subjective..
FIR analogue gets as close as possible to the ideal phase and magn. response, but trade-off is pre-ringing, significantly higher latency and CPU consumption.
Digital+ get very close to ideal magnitude response, but phase is uncompensated at all.
Digital+Phase corrects phase response to high degree, but it is bit worse in magnitude response then previous two and potentially can have issue with global phase shift.


Quote:
Originally Posted by FabienTDR View Post
BTW Vlad once wrote a quick and dirty article about the subject here: https://vladgsound.wordpress.com/201...alizers-draft/
That Vlad's classification is actually brilliant to me and I'm happy he put everything together to one page including links.. It can serve as great reference for threads like this one.

I'm also using oversampled EQs quite happily, incl. yours and Vlad's SlickEQ GE, which I like very much.

Frankly when I demoed Equilibrium last year, I haven't heard much of difference (if any) with real life material between standard modes and those, which were sometimes discussed here as favorite by someone (like some modes with insanely long FIR filters). Or it wasn't, I significantly preferred analogue phase to other oversampled EQs, I already have. All served very well to its clean EQ purposes.
What I like more about it was versatility and customization of UI (knobs, graph), curves and processing for different tasks.
So that time, it tried to defeat my GAS (or actually PAS).. and postponed it to later.. But as I'm reading through this thread.. maybe just for another round of measurements

Michal
Old 15th March 2016
  #41
Lives for gear
 
Hermetech Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by msmucr View Post
So if it is better or worse is probably bit subjective..
FIR analogue gets as close as possible to the ideal phase and magn. response, but trade-off is pre-ringing, significantly higher latency and CPU consumption.
Digital+ get very close to ideal magnitude response, but phase is uncompensated at all.
Digital+Phase corrects phase response to high degree, but it is bit worse in magnitude response then previous two and potentially can have issue with global phase shift.
Thanks Michal, it's stuff like this that is solid gold and probably should be in the manual. Whether one can hear the difference is gonna be solely material based, I am guessing.

I also love the TDR Slick EQ GE.
Old 8th March 2017
  #42
Gear Head
 
Mimieux's Avatar
 

Another two cents...

Quote:
Originally Posted by Conundra View Post
Aha!

https://www.kvraudio.com/forum/viewt...69602#p6069602

Looks like my memory failed me and it is an FIR filter used to correct the high end in Digital + mode, not an IIR as I wrongly remembered!

Cheers

Conundra
FWIW, here is another quote by Dave:

"It computes the spectral error between what Digital mode gives you, and the specified analogue curves do, and computes a minimum-phase correction filter of the specified size.

So, it is a fact that a digital EQ response cannot perfectly match an analogue EQ response simply because there IS a Nyquist. But you can get insanely close. The last 12 years of EQ research for me has been about getting absurdly close, and that's in EQuilibrium.

Digital+ compensation is a very simple idea - measure the difference between the analogue and digital responses and design an FIR filter to fix the error. Digital+ without phase just fixes the magnitude response. Digital+Phase fixes both phase and magnitude responses.

There should, in general, be virtually no improvement between Digital+ and regular IIR. It +SHOULD+ be extremely hard to measure and negligible in its difference. That means my maths is working well.

However, if I'm going to release what I consider the finest EQ built by human hands, it damn well needs to work 100% of the time. So Digital+ is there to catch any edge cases.
" (I've bolded the last sentence.)

-----------------

I tested various aspects of EQuilibrium last year (and I'm not claiming that my tests were necessarily perfect), and I noticed that IIR with Digital+ Compensation messes with the frequency of LPFs (the actual frequency of the filter tended to be higher than the numerical value displayed in EQuilibrium). I also noticed that IIR with Digital+ Compensation and Digital+ Phase distorts the shape of boosts in the very high frequencies (and increases the length of post-ringing). So I stick with plain IIR, which I think is intended to be the most practical IIR mode based on what Dave said in his quote above. The Digital+ options seem to be 'last resort' solutions for any unexpected (and presumably very uncommon) misbehaviour of EQuilibirum's IIR filter.

I've also noticed that plain IIR produces results that are extremely similar to FIR Minimum Phase (even if the FIR settings in EQuilibrium are maxed out). The only substantial difference between the two modes was an ever-so-slight distortion of an IIR LPF in the very high frequencies. So I generally think of plain IIR as a configuration that aims/tends towards FIR minimum phase (at max settings) but uses far less CPU, and I think of FIR minimum phase (at max settings) as a slightly more accurate version of plain IIR, which uses bucketloads more of CPU. In fact, more often than not, the phase distortion and post-ringing in IIR and FIR Minimum Phase (at max settings) seem to be virtually identical (but someone else may want to verify my findings by testing themselves). I'm always surprised by claims that FIR Minimum sounds vastly different to IIR because technically both filters produce extremely similar results.

Based on those tests, IIR, FIR Minimum Phase and FIR Linear Phase are the only modes that I use.

Last edited by Mimieux; 9th March 2017 at 02:54 AM..
New Reply Submit Thread to Facebook Facebook  Submit Thread to Twitter Twitter  Submit Thread to LinkedIn LinkedIn  Submit Thread to Google+ Google+  Submit Thread to Reddit Reddit 
 
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump