Digital EQ Fact & Myth.

rhythminmind · Post by **rhythminmind** » Sat Oct 04, 2008 7:04 am

Grappadura · Post by **Grappadura** » Sat Oct 04, 2008 7:31 am

Wow thx, thats a huge eyeopener! But there are still some EQs that are more effective in terms of steep cuts, right? Whats the secret behind that? I always thought the Cambridge EQ was the most desirable one because of this.

Crash · Post by **Crash** » Sat Oct 04, 2008 8:43 am

Hey Rythm,

seems like we're hanging out in all the same forums.

Thanks for posting your research!

ikeaboy · Post by **ikeaboy** » Sat Oct 04, 2008 9:51 am

Thats a real eye opener. So basically the price tag for a lot of these plugs is paying for the GUI????

I'd like to be surer about the differences between digital compressors. Do they not just use very similar look up tables to determine output from input. (I'm way out of my dpth here)

Crash · Post by **Crash** » Sat Oct 04, 2008 9:54 am

ORIGINAL: Creator

Glad to see you do the Waves SSL test. Analog mode only adds additional noise to the signal rather than changing the distortion modeling, however. It was ambiguous in your post whether you were stating that or something different.

Just a quick clarification on the Duende issue based on some of the things people mentioned earlier:
- All flavors of the SSL Duende products come with some form of SSL channel strip emulation, some with more in that area than others. The X-EQ is a separate add-on product based upon the technology in the pre-existing Algorithmix product "Blue" and emulates the curves from several different EQs rather than any form of distortion. This made a perfect candidate for the test performed here since there isn't any "x" in the equation for the "X-EQ".

I find a greater difference in "sound" between vintage emulations that model some level of saturation than I do between "all digital" EQs, like the Cambridge EQ, Sonalkis SV-517 EQ, Sonnox EQ, Sonitus EQ or Algorithmix Blue. The experience of using them can be quite quite different, especially as regards the organization of the GUI, and the curves they can easily do may vary (the Sonalksis can't go as narrow as the Sonnox for instance) .

But when it comes to "vintage models" part of the reason they tend to sound so different, even if they don't have saturation (which most of my favorites do) is that they have curves that don't work completely "logically". You turn on a band and suddenly a particular frequency is affected, regardless of what other frequency you are targeting. Or you have an inverse curve on another part of the spectrum from the one you are editing. These sorts of things can really help you gravitate towards certain sounds, if even if there is a huge overlap between what the different plug-ins can produce. That's one part of the reason why they are so useful as tools.

When it comes to EQs, I am happy to pay for the "x" in the equation and tend to do so more often than I'll pay for a high priced digital EQ. The way I see it, you really only need one good "pure digital" EQ plug-in, but in terms of which one is right for the user, it's practically like choosing a pair of shoes. Lots of options and it ends up being very personal.

I am very interested to hear what Massenburg has in mind though. If you've never talked to him before, I can tell you that it's hard not to get excited when he gets interested in something.

forge · Post by **forge** » Sat Oct 04, 2008 2:28 pm

ikeaboy wrote:So basically the price tag for a lot of these plugs is paying for the GUI????

no not at all. Actually what you are likely paying for most of all is the time and effort put into working out exactly what it is about specific EQs that makes them what they are, and one of the main things here is the curves, in particular assymetric bell curves

I highly recommend reading the Waves manuals, they really give a lot of detail - the Sonnox manuals do as well

in particular the Waves REQ manual goes into a lot of detail on what exactly it is that each of the curves are and why they are used

some really weird stuff happened with the pasting from the PDF and I got sick of fixing it, but you can probably download it from Waves

Craig Hutchinson from Manley Laboratories in the Waves REQ manual wrote:The most recent full-blown EQ in the Waves line up is the Renaissance EQ. Waves decided to make an analog of analog EQ. It was a research and opinion polling search to find what was magic in some of the vintage EQs and an exploration project to determine new and interesting approaches. A blend of old tools, fresh new tools, 48 bit precision, and an intuitive, simple interface all make this one special. It has some roots in the late Michael Gerzon's writings, describing the math to design some unusual EQs. Meir Shashoua wrote the code, fine tuned the program, accommodated dozens of requests and polished the GUI. Meir and this writer were email-chatting about the old Pultec EQs. Working at Manley Labs,I am pretty familiar with them. Rather than send a FAX or scan or draw curves, it seemed to be convenient to simulate each Pultec band using two or three Q10 bands and email these presets. The reaction was &'wow'. Gerzon's formulas were revisited and with the first prototype Mr. Seva heard the potential and contributed much
effort. Months later with good response from the working engineers who beta tested, it is a finished, distinctive processor.We hope you like it as much as they did.
You usually don't have to worry too much about levels with analog EQs but most digital EQs require some attention and optimizing of gains to get good re s o luti on and no cl i pp i n g. This is typ i c a lly the first thing that wi ll bug an analog en gi n eer. The answer for the RenEQ was to use a bit more DSP power and go for do u bl e - prec i s i on . This call ed for 48 bit prec i s i on ra t h er than the usual 24 bit math (on fixed - point DSP ch i p s ) . Th i s gives plen ty of h e ad room and a smoo t h er, less gra i ny qu a l i ty. Because of this incre a s ed math (DSP cycles) for
48 bit and more hors epower for some new curves and fe a tu re s ,t h ere is a limit of 6 stereo bands per instati ation (although newer DSP boa rds can handle more , dec i s i ons were made to keep back w a rd com p a ti bi l i ty ) .
On the other hand, e ach band is sign i f i c a n t ly more powerful and flex i ble than other EQs. Some of the frequency shaping curves are new by digital standards but old analog tricks. The Shelf curves
are inspired by the time tested Pultec EQP-1A's. With the old Pultecs, you have a one knob to boost lows and one to cut lows. Engineers favor using both knobs at the same time (secret). You might think that this
would just cancel out, as it would on most EQs, but it doesn't. Instead you get the full amount of boost in the deep lows, a steeper slope then a dip at the frequency or knee where it would normally be approaching
flat. The usual immediate reaction is that This is very FAT!.Why? On most equalizers, when you boost low shelves, you also boost the low mids and, to a smaller extent,the mids. When you boost lows on the RenEQ,
that is all you boost. A typical (single pole) shelf EQ generally has a rounded slope tapering at the top and bottom. At its straightest and steepest points it is about 4.5 dB per octave and gets lower. 4.5 dB/oct is gradual, (like a low Q on a bell) so a high shelf grabs a lot of high mids (esses) and mids (honk). These new slopes on the RenEQ range from 10 to 18 dB per octave (significantly steeper) depending on the Q control.
The dip at the knee is unusual and will probably hurt de-esser sales.
The “boost knob” toget h er with c ut knob of the Pu l tecs is con s i dered som ewhat con f u s i n g. Waves was abl e
to use the Q con trol in shel f m ode to ad just this curve from gen t l e to conven ti on a l t h ro u gh to Pu l tec . In case you haven't noti ced , the Q con trol is a de ad knob in shel f or filter modes on other EQs and most EQs
c a n 't can swi tch bet ween bell ,s h el f and filter mode s .Most have ded i c a ted bell s , 2 shelves and 2 filters and if
you want “a ll bell s”, or 2 high shelve s , or som ething different from “s t a n d a rd ”, yo u’re screwed . Ye s ,Wave s
a ll ows a little more freedom and both the Q10 and RenEQ has two or three kinds of c u rves for every band.
For the first time we know of, this steeper type of curve has been applied to the high shelf. Finally, we have a
new shelf EQ, that when we want to boost the upper highs, we don’t end up with nasty esses.We get real
air”and “sweetness’ instead of harshness and that hard edgy sound. Because you now have an active Q con-
Renaissance Equalizer Plug-in Manual 23
trol in the shelf modes, it allows a whole new level in “finessing” the tone. It might be considered a little
breakthrough, like the original “Parametric EQ” which added an independent Q control to the bell curves.
Speaking of bell curves... These are a little different too.Waves calls these “Asymmetric Bells”. The boost
shape is conventional but the cut shape may be new to you. For a pleasant change, it is not a mirror image
of the boost. There have been some similar relatively unknown analog counterparts, but none that were
available to the pro engineers before this.Waves gave their field testers a choice between conventional bells,
Asymmetric Bells or both. Apparently, the overwhelming choice was the Asymmetric Bells (only). It seems
that, at any more than a few dB of cut, we tend to prefer narrower Qs. This is quite the opposite of boosts,
where we usually prefer wider Qs. This EQ does this for you semi-automatically. It also allows you to get
deeper cuts without messing up as much frequency area that you wanted to keep. If you like wide gentle
cuts the RenEQ can still be unusually wide and smooth too.
There will always be a few who say “the advantage of a symmetric bell is that it can be used like an undo
button”. You can restore some overdone EQ by setting up the opposite curve.We can’t dispute that, and the
Q10 does that very, very well. However, this seems to be a reasonably rare need.Most engineers lean towards
approaching EQ conservatively, and most often, find themselves boosting a bit more where they boosted
before (or vice-versa) or just fine tweaking a recorded track with whatever works best. It seems most guys
don’t make those kind of mistakes or require “un-EQing” to be THE prime concern. They would rather
have an EQ that simply sounds magnificent. Some conservatives may argue that “all other EQs are symmetrical,
and its probably the only real way to do it and must be the best”.When have you ever been given a
choice? Working engineers chose these Asymmetric Cuts because they really like how this EQ sounds. Now
you have a choice. You can use the Q10 with symmetric bells or the RenEQ with these new cut shapes. You
should sit down and compare them, get a rough idea of what Qs you like for little boosts, big boosts, little
cuts and deeper cuts. Check it out.We think it sounds more “acoustic” as opposed to “electronic”.
We borrowed these cut curves from nature and are steadily paying “her” royalties.When sound reflects off a
surface, it bounces back and interacts with the original sound and creates a harmonic series of boosts and
cuts. The dips are caused by the two sounds being out-of-phase and canceling at some frequencies. Actually,
they rarely totally cancel out because the reflection is not the same volume. If the reflection (or second
wave) has a similar frequency response we expect a comb filter which is a series of boosts and dips. Mostly
when you see “real life” responses, you can clearly see only a few dips and they look like the “Asymmetric
Cuts” on the RenEQ and not the shapes found on other EQs. This makes the RenEQ a “natural”as a notch
filter.We use notch filters to remove pitched noises like hum or annoying instrument resonance’s. Another
thing to note is that lower Q values are needed to get a good narrow notch and this reduces resonance and
ringing in the time domain.A higher Q has a longer decay time. So, you got yer old magic in the shelves,
natural bell curves, and more control and versatility.What more do you need? Did we forget the filters?

and from the Sonnox manual:

6 EQ types included in the Oxford Plug-in.
Programme equalisers have expanded, beyond their original use as distance correction devices for film and vision, into highly creative tools that represent a leading part of the sound engineer’s artistic palette. A great many EQ designs have been developed over the years that have been attributed with qualities that lend themselves to particular uses and sounds. The Oxford EQ plug-in is designed to be flexible enough to address as many of these generic types as possible from a single application, by presenting a variety of types to the user. The following pages are presented as a general explanation of many of the factors that affect EQ performance and to illustrate how we have addressed these issues with the Oxford EQ plug-in.
Many types of EQ exist with many areas where they differ. One of the most important areas is the issue of control ranges and interaction. Whilst it is true that with a parametric unit with continuous controls (i.e. not quantised) any response could be obtained by matching their curves, many of the popular EQs have control dependencies that err towards specific application. One of the main areas where EQs differ is Gain / Q dependency. Most analogue EQ has Gain / Q dependency as a result of the circuits used. This factor can greatly affect the artistic style that an EQ presents by facilitating certain parameter settings and encouraging particular uses when the unit is operated.
In the Oxford EQ plug-in we have covered this situation by providing 3 different styles of EQ that take account of Gain / Q dependency as well as overall control ranges. The following section describes these options.

......

EQ type 1....This style has minimal Gain / Q dependency, smaller amounts of boost or cut still have relatively high Q and it is therefore precise and well defined in use.
However it is sometimes difficult to obtain overall EQ fill on combined sources and subtle EQ on vocals and the like, as the user needs to adjust the Q control to maintain an effect when the gain is changed. Failure to understand this fact has often added to the reputation of this type of EQ for sounding ‘hard’ or ‘harsh’. However, because the user retains separate control of all its parameters, this EQ is still the most flexible for users that have the time and patience to spend when using it.
It is most like the original 4000 series SSLs, and other ‘clinical’ styles of EQ that became popular in the 1980’s.

EQ type 2......etc

point being, no, it's definitely not the just GUI you are paying for

@Rhythminmind - I see what you are saying, but what is the point you are really making? sure it is probably possible to recreate different EQs with any digital EQ + some saturation etc, but in reality re-creating specifics to such a degree would be quite an undertaking really as the designers of these plug-ins and Craig Hutchinson demonstrated in the above pasting from the REQ manual - they didn't just knock up something that sounds like it, they spent a long time working out what to do and why

having said that, the vast portion of the time I'm prefectly happy with Live's EQ8, and I use REQ mainly for when I want to be really surgical with hi/low pass filters as the Q has no gain boost, but occasionally I do use it for it's character

funky shit · Post by **funky shit** » Sat Oct 04, 2008 2:38 pm

I wish i knew wat all this shit meant

forge · Post by **forge** » Sat Oct 04, 2008 2:49 pm

Crash wrote:
ORIGINAL: Creator

....

But when it comes to "vintage models" part of the reason they tend to sound so different, even if they don't have saturation (which most of my favorites do) is that they have curves that don't work completely "logically". ....

Oh I see my point was already made here by Crash

Crash wrote:
I am very interested to hear what Massenburg has in mind though. If you've never talked to him before, I can tell you that it's hard not to get excited when he gets interested in something.

I'd be interested to read the original post, where is it?

j2j · Post by **j2j** » Sat Oct 04, 2008 2:56 pm

Does this place really need to mirror KVR?

If it sounds different, it is different.

1 + 1 = 2
1 + 1 + x = x

and only 2 if x = 2

Yes, maybe some of the algorithms are the same. I don't know, I don't do C++

But then again no, cause they all sound and act very differently, and that ought to be mentioned.

Never underestimate the value of X.
Cheers

j2j · Post by **j2j** » Sat Oct 04, 2008 2:57 pm

forge wrote: I'd be interested to read the original post, where is it?

http://www.kvraudio.com/forum/viewtopic.php?t=229329

forge · Post by **forge** » Sat Oct 04, 2008 2:59 pm

j2j wrote:
forge wrote: I'd be interested to read the original post, where is it?
http://www.kvraudio.com/forum/viewtopic.php?t=229329

thanks

I found the one on Gearslut

Crash · Post by **Crash** » Sat Oct 04, 2008 3:12 pm

forge wrote:having said that, the vast portion of the time I'm prefectly happy with Live's EQ8, and I use REQ mainly for when I want to be really surgical with hi/low pass filters as the Q has no gain boost, but occasionally I do use it for it's character

I may have misunderstood the concept of a hi-pass filter, but this is not what comes into my mind:

Fortunately it's mostly happening in inaudible dB ranges.

forge · Post by **forge** » Sat Oct 04, 2008 3:15 pm

curious!

although I can't say I'm terribly worried about anything that low

{edit}remember, the SNR for CD os somthing like 96dB - what you are showing there is well below even that, for tape it was bout -55db!

24-bit digital audio has a theoretical maximum dynamic range of 144 dB

http://en.wikipedia.org/wiki/Audio_bit_depth

out of interest, was that in Hi-Q mode, if not does it make a difference?

Crash · Post by **Crash** » Sat Oct 04, 2008 3:28 pm

forge wrote:out of interest, was that in Hi-Q mode, if not does it make a difference?

Same settings in Hi-Q:

To check wether it's just Hi-Q's oversampling at work I checked again without Hi-Q at 176.4 kHz (=4 x 44.1 original rate of the clip).

OUCH!

j2j · Post by **j2j** » Sat Oct 04, 2008 3:29 pm

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:

http://citeseerx.ist.psu.edu/viewdoc/summary?doi=1 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?