Digital EQ Fact & Myth.

[bmanic]

A lot of "classic" analogue EQs do not cut nor boost exactly like it says on the label of the knob/slider. ... That single knob might boost at 3kHz, dip a little at 200Hz and perhaps lift the extreme high end a bit at 16kHz.

i guess, that is what is meant with 'properly designed'. in my opinion, 'properly designed' should mean that it does what it says.

Of course but that is not the point. Just like some guitar stomp boxes are "happy accidents" these old time EQs have become sought after for a particular sound/behavior so if a developer says they are emulating the unit it should do the emulation as close to 100% as possible, right?

Cheers!
bManic

[Music Engineer]

@Rock:
yes, the +X is not taken to be literally (in sense that the same effect could be achieved by chaining some other effect behind the EQ). more like 'EQ with some artifacts'

[Music Engineer]

in my opinion, 'properly designed' should mean that it does what it says.

Of course but that is not the point.

it was not your point, but mine it was. and i think (more importantly), the OP's as well

Just like some guitar stomp boxes are "happy accidents" these old time EQs have become sought after for a particular sound/behavior so if a developer says they are emulating the unit it should do the emulation as close to 100% as possible, right?

sure, if some unit says to exactly emulate some analog hardware, then it should do exactly that. ...and i love the sound of guitar-amps btw. nonetheless, if i want EQ then i want EQ and not all kinds of stuff extra

[bmanic]

Emulation plugins for the most part are Hype. But you might find the interface more creative, Or you might like that type of saturation. But in reality you would be better off with using a fully variable saturater & an EQ.

As you can see, the original poster clearly states Emulation plugins and basically says you can use a saturation unit and a normal biquad based EQ to do everything you need.

This is in my opinion not true so I stand by my words and opinion in that I think you lost the plot a bit.

The way a hardware box reacts to your tweaking is very much a big part of why you chose that particular EQ in the first place. The quirkiness makes you do certain things and come to certain results. Sure you can pretty much "match" the sound of that with a saturation plugin and a basic biquad but to do that you need a reference, the original performance/tweaking of the vintage EQ! See the dilemma?

Cheers!
bManic

[Christian Budde]

Of course but that is not the point. Just like some guitar stomp boxes are "happy accidents" [...]

I once had the chance to talk with a developer during Musikmesse from the 'old days'. He really knows a lot about electronic parts and how to create circuits properly. Despite all his knowledge they wasn't able to completely understand the circuits they built. There wasn't simulation programs available and they only could do a rough estimation of how the circuit might sound like after the assembly. However, they built in some variable resistors here and there, so that they were able to 'tweak' their prototype.

Then they hired some people like a guitar player for example to play with the device they build (while changing the values of the resistors). They did so until the result got 'better'. Probably quite a subjective perspective, but sometimes the only way how it worked. They probably constructed several prototypes, because sometimes the best sounding constellation had some bad side effects (like noise or unwanted distortion). As he told me, they tried their best to achieve a good sound with less unwanted side effects.
In the end they came up with a probably 'overengineered' device (whatever that means).

From todays point of view (with all the simulation programs at hand) it most likely wasn't the optimum they could theoretically achieve, but the best solution they found (as mathematics probably would say they found a local minimum/maximum).

In case of EQs the filters were most likely minimum phase filters or at least close to minimum phase filters. Or at least a network of minimum phase filters. Due to the circuit I guess the result must not necessarily have been minimum phase, but could impossible have been linear phase either (as it would imply taking a look into the future or having a delay line, which was quite difficult to achieve in former days).

For the use in compressors they might have been constructed a Bessel filter, which has linear phase like properties in the transition band. However, I guess the filter ladder must have also lead to a lot of noise and emphasized the distortion. So they probably modified it to be more stable (it actually should have been stable, creating distortion, where a digital filter would prefer to 'blow up').

In the end they might have had a device that wasn't exactly like they had it in mind, but all the flaws and issues they filled up with their creativity to turn the issues into features.

Speaking about noise and distortion, the person I was talking with was a bit puzzled. He told me, that if it was only up to them they would have never came to the idea to put noise or distortion in a device by purpose.

Yesterday I was having a discussion with a friend of mine of exactly that topic. I ask him to think about the difference of writing a letter rather then writing an email. Even if you are probably delivering the same message it's still a difference between both types. While writing a letter probably has the problem for the reader that he might not be able to identify every single word (because of the handwritten font, probably identical to noise here). On the other hand the reader might spot more information by the way the words are written. Also, even if the original author of the letter will write it again, it will never be exactly the same. It's simply hand made and unique.

On the other hand, the email is more easy to handle. You can simply make a copy of it and send it around the world. However, since less work is involved the writer might not think about every single word as if he would probably do by writing a handwritten letter. Also note that the workflow is probably more easy (having an spell checker integrated into the mail program). Or you can use a text you already written once.

Comparing both mediums, you can for sure transmit the same message. Though I'd guess in 10 years you would rather remember that handwritten letter then the email (even if both deliver the same message). Well after that period the original letter might not be readable anymore, but you probably still have the email.

To achieve the best of both worlds you could also think of writing a letter, scan it and send it. This is probably equal to tools like Nebula. You actually capture the whole atmosphere and transfer it into the digital world. Unfortunately your still not able then to use the spell-correction to correct a wrong word with this approach.
Other ways to achieve a mixture would be to not only send an email with plain text, but with formated words. Probably add a scanned picture to add more atmosphere here and there. Still it would never be the same as writing a letter, but this way you can create something even superior combining both worlds. It's probably just harder to add the creativity, that might have come to your mind while actually writing the letter.

It might not be the perfect comparison here and I almost forgot, what the original topic was all about, but I hope to open your mind with this post and give you something to think about (without adding too much of my opinion to it).

Christian

[bduffy]

That's an amazing analogy, comparing the handwritten letter and the email. Very interesting, Christian.

[justin3am]

Some great info in this thread, it's a topic I'm very interested in. I find that, depending on the material I'm drawn to work with specific tools... I never really though about it but when I'm EQing on a per track/ per instrument basis I tend to grab EQs that don't have a filter curve display and when I'm EQing a buss I generally grab the ones that provide as much feedback as possible.

I'm constantly questioning weather I can hear a difference between my tools or if it's merely psychological (especially when differences seem to be subtle), I'll have to investigate further the next time I do a mix.

One thing I've always wondered, what makes one filter response more pleasing than another to me? I suppose that it's based mostly on A) personal preference (duh!) B) how I perceive the filter is interacting with the levels of frequencies that make up the character of a sound. Deeper than that though, how did I develop my sense of a pleasant interaction v. an unpleasant interaction. I assume that my relationship with the material I'm working on has a lot to do with how respond to the effects of a processor (e.g. if I'm working on my own stuff v. someone else's material or if I'm trying to repair poor quality audio v. sweetening well recorded audio), do you suppose that when I'm displeased (or pleased for that matter) with a processor it may have more to do with my mindset than the actual effect it has?

There are so many variables that contribute to how we perceive sounds... I find that detailed metering and such is very useful but sometimes it almost seems counterintuitive because it's like I'm hearing something different from what is being displayed (in which case I turn off the displays and play by ear)!

Any thoughts?

[Zaphod (giancarlo)]

To achieve the best of both worlds you could also think of writing a letter, scan it and send it. This is probably equal to tools like Nebula. You actually capture the whole atmosphere and transfer it into the digital world. Unfortunately your still not able then to use the spell-correction to correct a wrong word with this approach.
Other ways to achieve a mixture would be to not only send an email with plain text, but with formated words. Probably add a scanned picture to add more atmosphere here and there. Still it would never be the same as writing a letter, but this way you can create something even superior combining both worlds. It's probably just harder to add the creativity, that might have come to your mind while actually writing the letter.

yes, it's the same difference between samplers and virtual synthesizers. But even today I work with a single virtual synthesizer (virus), and ONLY samplers or sample-based instruments. Eveything else is turned off.
Don't ask me the reason why, but that "scan and send it" could be creative enough with scissors and glue if you are doing patchwork. The extra limitation brings new creativity.

In any case, I agree with niklas A LOT. When I'm looking for an api emulation, I want EXACTLY that. Not more, not less. I'm not looking for a powerful instrument with new added characteristics, sound, modulations, pitch shifting or whatever, spectrum analyzer and spectrum match, real time compression or whatever. For that I have already 10000000 free plugins around.

[Christian Budde]

Don't ask me the reason why, but that "scan and send it" could be creative enough with scissors and glue if you are doing patchwork. The extra limitation brings new creativity.

I hope not to sound like judging somewhere, so whatever works for you is just perfect. To keep in the analogy and speaking for myself, I'm more that email guy, but more because I'm probably to lazy and can't really remember of all the formal things you used to write letters (like putting a date on every page and leaving a margin and stuff like that). I'm quite used to write emails every day as my mother is used to write letters (she doesn not really like computers at all, but that's another story).
What I want to say with it is, that some people prefer the 'traditional' way, while others would probably use the 'modern' way. An emulation of EQs is probably somewhere in between.

Not discussing about the fact how EQs actually differ, I just want to point out here that probably "older" hardware EQs with probably some limitations have their charm (regardless if and how they differ). Even if the sound might be equal (which would be an identical message in my analogy), the workflow is still different and in my opinion probably even more important then matching EQ curves.

As said in the other thread I have done a lot of research in case of EQs and I think it's possible to match nearly every EQ curve to a point where the difference is probably impossible to hear, but the workflow will make the difference. Some prefer the traditional way, others the modern way. Both have their advantages and disadvantages.

But comming back to the original topic. In my opinion most typical digital EQs actually are equal (to a certain point) and you probably could replace one by another. You probably don't even need to buy any other EQ but the build-in host EQ. However, it would be wrong to say, that all EQs are equal only because they might result in an equal frequency response. As I tried to show above it's also the workflow. Some EQs affect the sound 'just right' by dialing a few dials (while the curve itself probably can be achieved by any other EQ as well). Others need a lot of tweaking to get to the desired results. It might be a different frequency-gain-bandwidth dependency or just a good initial preset. Sometimes such effects have more impact then you probably think.

I once made a test here at KVR to evaluate whether a difference between floating point and fixed point calculation in EQs can be heard. I created a plugin with several selectors (probably 15 or so) and I told the people, that at least one of those algorithms is floating point and at least one was fixed point and that they should judge what they are. In the end only 2 of the filters presented were fixed point, while the rest were all floating point. I have no idea about the competence of persons joining the evaluation, but there were quite interesting responses. Some persons even 'spotted' a third algorithm. None of the listeners came even close to the results. It seems that even in case the algorithm is completely identical some persons might be fooled only by their brain. Not to imagine if I would have changed the color of the GUI or the knob positions (which surely means a different workflow).

Just some more comments to my post earlier,

Christian

[docdued]

To get back to the original question: I've tested dozens of software EQs, and they fall into two categories: those that use the standard biquad implementation and those that don't. More than 90 percent fall into the first category (the probability to be in this group is increasing with how many times the word "analogue" is mentioned in the advertisement ), and for these cases it's really "only" the interface, the workflow and the CPU consumption that make the difference (plus they might work with or without oversampling, which leads to different behaviour near the Nyquist frequency). However, there ARE other EQs: linear/variable phase EQs (or generally FIR EQs) as well as those that use some variant of the Orfanidis filter for peaking. And those that add saturation, of course.

[Zaphod (giancarlo)]

...

well, for some reason this world today is sample based. Even for 3d games (where everything is calculated in real time, for example lights) the core object is still the texture, which is... a sample. Maybe there will be a future (far future) when everything will be calculated in real-time, emulating every suble nuance of reality (using complex formulas and maths). I'm asking to myself if it worths and if it is possible. Maybe it will be.
If you need a digital equaliser a biquad implementation with a good oversampling stage is good enough. If you need a good emulation you could derive the electric schema, and it will be light on cpu. But if you need a close emulation you can't beat sampling, period. There is no reason for not using it, apart the cpu load, which today is coming more and more neglegible. From the user perspective, you know that it's simply a matter of control, but you know that control is a consequence of the implementation, not of the technology used. You could sample a photo in real time, and derive a vectoralization: technologies could be mixed as you want. But before... you need to build a sample library.
Think to the powerness of google earth. Before they sampled the whole earth. They didn't measure everything. They sampled from the space.

Speaking about your analogy, you don't need to sample the whole document. Sample just images, and add your text inside your wordprocessor (the sequencer). There are so many alternatives around. When you need a photo... well, take a photo.

[jupiter8]

If you want to make a lot of friends fast,post this over at gearslutz. They're going to love you,i promise.

Well i have & all i can say is KVR is surprisingly more hostile.

As the mighty Christopher walken has said. "It was a horse deal with it"

You did ? Where ? I must have missed it.

Actually i've seen quite a few threads that supports your findings. That Lagerfelt fella had a thread where he emulated some eq plugin using Logics builtin eq.

And i remember the Sonalksis (i think) guys saying something like "In a standard eq implementation you have 3 degrees of freedom plus distorsion" which means there isn't much room for any differences.

[Arksun]

I thought i would share my research as of late.

I'm going to make a broad statement that seems to cause some to freakout. Read the whole post before replying.

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

Don't just take my word for it, listen to the developers.

Read this from algorithmix Algorithmix

The truth is that with a properly designed, fully parametric analytic PEQ, every amplitude and phase characteristic of any other equalizer setup can be recreated. Of course, the contribution of distortions to the specific sound of a particular analog equalizer caused by the respective electronics has to be considered. If the distortions are 'good', they may make certain applications sound better.

So whats this mean?

Emulation plugins for the most part are Hype. But you might find the interface more creative, Or you might like that type of saturation. But in reality you would be better off with using a fully variable saturater & an EQ.
But one thing most emulation plugins do offer is an included compressor. Unlike EQ's compressors come in many shapes and forms. & tools like the URS CSP with it's input stage really can change a compressors quality.

In the end the quality variables are down to internal resolution, oversampling capabilities, & so on. There is no magic in the emulation.

Interesting findings, though perhaps slightly flawed for several reasons:

1) You've quoted Algorithmix and taken it entirely out of context. That paragraph is stating that its possible to recreate any kind of eq with a properly implemented designed analytic eq. That is not the same as saying all digital eq's already out there can and do perfectly recreate each other.

2) Yes, you're right a lot of digital eq plugins out there do share the same algorithms, but not ALL are equal. Comparing just two, or even 4 or 8 of your favourite eq's, one cannot draw the conclusion that every digital eq plugin made can recreate all the others with a little tweaking.

3) You're talking about them all being the same, whilst at the same time pointing out the differences, even with those that use the same kind of algorithms for the curve shapes, with things like saturation/distortion, oversampling, linear phase etc. I could build a car from parts to get me from A-B, but it would take me 6 months to get to my destination instead of 15 minutes if I just had the car already. This ties in with the thread about plugin simplicity. A big part of it is about getting the tool to do what we want in a musical fashion as quickly as possible. If one eq has curves and shapes that are immediately more musical or feel more transparent than another, who wants to spend an hour desperately trying to tweak another eq to match those curves. And if you've only bought that 'other' eq and not the very musical one, you have no reference to match to anyways!.

4) Do freq analysis and phase tests take into account phenomina such as bell ringing on tighter Q's and general transient response when applied to complex material?.

Why is it that when I've compared different linear phase eq's and done null tests on a low shelf , I hear differences in the artifacts left behind throughout the entire frequency range (PlParEQ came out top with no artifacts at all)

In the case of the Eiosis AirEQ I really do notice a difference to its sound to other eq plugins I've tried (and I've tried to match the curves). The way it affects the stereo soundstage is different to say the Sonalksis EQ.

But in the end its not what's under the hood, or the brand name, or the cost that matters. All that matters is that the eq allows you to achieve the desired goal as painlessly and quickly as possible, without trying to tweak constantly to the point you just end up frustrated and it saps your creativity.

So yes, its a fact that 'many' eq plugins employ the same algorithms and could recreate each other. But its a 'myth' that they're all the same and that you can easily recreate any one from another in as fast and productive way

[SWAN808]

Arent we underestimating the importance of X here?

[Music Engineer]

Arent we underestimating the importance of X here?

X is certainly often important, but i prefer to add that separately.