Zero Delay Feedback Filter (How to test if your synth has a )- Xils-Lab White Paper -

[Frantz]

There are plenty of workhorse synths that sound very good with moderate cpu already. We have already excellent Toyotas.

Yes, my favorite Toyota is the Korg Polysix. I can run as many as I want without straining the CPU. But this Toyota is old, rusty, and rattles on the highway. The problem is I can't find a newer, better replacement.

And how would you know?

I am rich (in imagination).

so you settle for second best to save on cpu ?

Sure. Do you have the absolute best speakers, amplifier, TV, car, house, etc? Maximizing quality without considering cost is not practical. Life is a series of cost-benefit analyses.


Sorry for the hijack. Back to 0df, non-linear, trapezoidal, z-plane, Nyquist whatchamacallits.

[Lotuzia]

I'm a guitartist and I'll tell you some guitars suck, they go out of tune easily, have warped necks, bad workmanship, cheap parts. Same with software, some plugins suck: they crash, have poorly designed menus, broken preset systems, frustrating copy protection, terrible support from the deveopers. Crap exists, it is not all subjective opinion.

Oh you answered before I edited my post because it was poorly written.

So I will try to develop a bit here because so that your answer is well understood by some eventual readers.

Concerning Cpu stress, I took the example of FM synths, to change a bit, while not giving names : I prefer an old FM synth, that have a very low cpu stress, to some more modern ones, that have a major stress. Why ? because just for me I think the modules, their interconnections, the workflow, the features, the enveloppes, the way operators are connected, the design, the ideas of what can make a nice musical instrument, are more to my taste than in some other ones, and, in the end I prefer the instruments I can create with this good old synthesizer.

Then I took a comparison with guitarists, hence your answer, and say that its only in the small software instruments world than some people compare instruments with better/worse comparatives. My idea is that Guitar players wont say : Fender are better the Gibsons, or Takamines;. They will just say : i prefer to play on Fender, etc, its a matter of personal taste and that's it. ( so now people can understand your answer )

I know some pianists who prefer to play on this or that piano, but here again I've never heard someone saying : My Bosendorfer is better than your Steinway, my Rhodes than your Wurli.

Cpu cost again ; I prefer the old ImpOSCar to some more modern synths with demanding CPU needs. Its old ( 2005/6 ?) but to me it has a particular sound I like.

Then I also wonder about all the music made with the Imp' or such dated instruments. Its was good, it is still good.

So as a conclusion, a Ferrari is a synth that can be part of good music. Whether it is old, eats or not cpu, I dont give to that more attention than required, because a musical instruments is much more than a pile of technical specs.

In the computer world, however, there are really Toyotas, Ferrari and Trabants ( And sadly mine are merely Toyotas also ) so I can perfectly understand why people make the link between their Toyota computers and the instruments thatthey will be allowed to play with relative comfort.

-------________

As for your answer, well, I was talking about choices between merely mature musical instruments. ( But tbh 'poorly designed menus' as first criteria for a synth, its far from the idea I have from a musical instrument )

[AdmiralQuality]

My idea is that Guitar players wont say : Fender are better the Gibsons, or Takamines;. They will just say : i prefer to play on Fender, etc, its a matter of personal taste and that's it. ( so now people can understand your answer )

[Lotuzia]

+ Sorry to AD for the same reason / I pushed the post button instead of the preview post that I usually use to try to erase at least the most ugly parts of my "franglais" so now you can answer, its a very similar post.

[braj]

Hahaha! BwaHAHAHAHA! Yeah right! That is like saying Ford guys like Chevys!

"Disco sucks and so does soul, all that's left is rock and roll!"

Have you ever even MET any guitarists?

Anyhow thanks for the laugh!

[Lotuzia]

My idea is that Guitar players wont say : Fender are better the Gibsons, or Takamines;. They will just say : i prefer to play on Fender, etc, its a matter of personal taste and that's it. ( so now people can understand your answer )

Well I've played, or talked, with tons of guitar players, and fortunately no one what like that.

[AdmiralQuality]

Hahaha! BwaHAHAHAHA! Yeah right! That is like saying Ford guys like Chevys!

"Disco sucks and so does soul, all that's left is rock and roll!"

Have you ever even MET any guitarists?

Anyhow thanks for the laugh!

I know, eh? Who's ever met an opinionated guitarist? They're all so fair minded!

[Lotuzia]

Hahaha! BwaHAHAHAHA! Yeah right! That is like saying Ford guys like Chevys!

"Disco sucks and so does soul, all that's left is rock and roll!"

Have you ever even MET any guitarists?

Anyhow thanks for the laugh!

I replied to your first post with courtesy, now this is my answer.

A lot of guitarists actually. Fortunately no one was as arrogant as you seem to be. ( but pro guitarists are very rarely arrogant tbh )

I leave you to your nice world populated with people hating those who dare to think in a different way than their own clan. In a way, you behave just like them. My world is obviously different. Oh and its populated with real people too, just not the same than yours, and certainly not connecting with each other in the same manner : You might call that respect, if this word makes any sense for you.

best,

LtZ

[braj]

Hahaha! BwaHAHAHAHA! Yeah right! That is like saying Ford guys like Chevys!

"Disco sucks and so does soul, all that's left is rock and roll!"

Have you ever even MET any guitarists?

Anyhow thanks for the laugh!

I replied to your first post with courtesy, now this is my answer.

A lot of guitarists actually. Fortunately no one was as arrogant as you seem to be. ( but pro guitarists are very rarely arrogant tbh )

I leave you to your nice world populated with people hating those who dare to think in a different way than their own clan. In a way, you behave just like them. My world is obviously different. Oh and its populated with real people too, just not the same than yours, and certainly not connecting with each other in the same manner : You might call that respect, if this word makes any sense for you.

best,

LtZ

Hey, you just disrespected ME more by your post than I ever did to you. If you don't see that well, what can I do? You obviously misunderstand much.

[AdmiralQuality]

Hahaha! BwaHAHAHAHA! Yeah right! That is like saying Ford guys like Chevys!

"Disco sucks and so does soul, all that's left is rock and roll!"

Have you ever even MET any guitarists?

Anyhow thanks for the laugh!

I replied to your first post with courtesy, now this is my answer.

A lot of guitarists actually. Fortunately no one was as arrogant as you seem to be. ( but pro guitarists are very rarely arrogant tbh )

I leave you to your nice world populated with people hating those who dare to think in a different way than their own clan. In a way, you behave just like them. My world is obviously different. Oh and its populated with real people too, just not the same than yours, and certainly not connecting with each other in the same manner : You might call that respect, if this word makes any sense for you.

best,

LtZ

So, anyone who dares to disagree with you is a "hater". All opinions are valid, even idiotic ones. And we gather here to accept your wisdom without daring to question or doubt.

I'm done here. Toodles.

[Kriminal]

nevermind

[Kriminal]

My world is obviously different.

that would be The Land Of Make Believe then

[xavier]

just wanted to point out that every 24db series lossy integrator "ladder" filter i've tested fails this test. (we're talking analog, electronic here. definitely 'zero delay'.) none of them have perfectly flat feedback/hz usually because of filters in the feedback path but also because of various issues like OTAs which don't all respond with 100% accuracy across their entire range. for example the coefficient for one integrator may be different than the other three which will influence the amount of feedback required to maintain an exact peak level.

of course nothing we didn't already know. test is 100% bullshit and seems to be based upon the assumptions of an ignorant. for example the apple falls due to the unseen hand of god pushing in a single downward direction. only this no longer makes sense when the earth isn't flat or once your universe is no longer ptolemaic.

I feel strange that a lot of you are posting statements without any images, videos, explanations. When I asked, the only answer I got is, "no need proof of evidences, I'm talking logic" (apart AQ who brought something constructive, thanks to him).

When you are saying "fails" this test, doesn't it means that the resonance goes from Mid setting to self-oscillate settings like in the posted video test ? Or just shows a little variation, due to the electronics ?

If it is the first case, than please, post your examples, I want to lean, !!!!

If it is the second case (for the instance the resonance drop-down in the bottom of the spectrum, bellow around 100 Hz) I think you didn't read what I said/posted/explained several time.

SO An Other time : If the test shows that the digital filter has the same Frequency/Resonance behaviour than your electronic analogue filter, then you can say that your digital filter s correctly managing the internal delay(s).

In the same way, testing an digital filter, which is "emulating a analogue filter" and getting some frequency/resonance link. The question every one should ask is "is the link shows for this digital filter is the same than its analogue model ?".
Of course this implies that the developer "kindly" tell which model he used.
On the contrary, it too simple to publish a non accurate filter emulation and said "hey, what you are seeing is not because our filter is not accurate, it is because the model is like that",without publishing any proof.

Here is a plot of Moog-Ladder analysis simulation (nothing digital as the frequency range can show, but a theoretical analysis of the analogue schematic, so "definitely 'zero delay'" )

Extracted from "Analysis of the Mood Transistor Ladder" by T.E Stinchcombe, an analogue synthesizers fan.



Even if this analysis is not 100% accurate, it doesn't show the same behaviour you can see/hear in the test video (resonance going from mid-setting to self-oscillate settings), only a very little variation.

So saying "This test is incomplete, it does not take into account all the cases, etc ... ", it's ok and I'm fully opened to discuss.

But saying "this test is 100% bullshit based on assumptions of an ignorant", then we can't really discuss and my feeling is that this assumption is done by a Digital Signal Processing ignorant.

Best retards
Xavier

[aciddose]

actually the simulation isn't "definitely zero delay".

first of all, it's constructed using the same elements urs is using in diva (according to him, i have never looked at the code.)

secondly and more importantly, you're running the simulation without taking into account many factors which influence the real-world circuit. half of these things are where you're assuming the devices act as their ideal models, and the other half are that you're forgetting to simulate many effects that are outside the scope of the basic simulation.

these things are not automatically managed by simulation software. even good simulation software requires manual modelling by the user to create accurate contexts for simulation of circuits.

you're also making a huge and ridiculous generalization when you assume feedback will increase as frequency increases. this is actually the opposite of the result i get in software - but that is due of course to the fact that my method (of four stage series lossy integrator and global feedback) over-compensates in it's approximation. as it turns out running two 12db filters in series with compensation is actually far less expensive than running one 24db filter.

so you're talking about a single implementation which is uncompensated or under-compensated.

in your simulation, try modifying the transistor VBE in the ladder in monte-carlo analysis and graph in 3d the variations of the graph you have posted. you'll find you can get a wide range of effects.

i've tested my own circuits on the bench. ota, biased transistor, diode. none of these have flat peak levels as you adjust frequency unless the parameters of the circuit are ideal. i've also tested a few variations in software with the same results. you can over-compensate, under-compensate and you can also easily get better peak normalization than the analog implementations if you put enough effort into the compensation.

the fact is this test is simply invalid. it makes ridiculous assumptions and it measures a symptom with a myriad of potential causes without the slightest accounting of those.

in reality the only way you can measure whether a filter is zero-delay or not is by phase graphing.

your test can be fooled by proper compensation of feedback, which you apparently assume is impossible.

don't go around calling people ignorant when you have no idea about how to produce the content of the paper you're posting. you should be able to produce this yourself and you should already know about all the effects i'm talking about. i would assume you did if your arguments made any sense but as they seem so outright foolish i'm forced to assume otherwise.

[ariston]

But saying "this test is 100% bullshit based on assumptions of an ignorant", then we can't really discuss and my feeling is that this assumption is done by a Digital Signal Processing ignorant.

So, we're basically back to square one here. To sum up this lovely thread (imo imo imo, of course, else "someone" begins to accuse me of speaking for everyone else again):

We have two issues here. One is the technical question. As the technical discussion wore on, it became clear that this was a topic for experts. A lot of tech-speak flying around, with niceties that basically leave all the non-experts out in the cold. Which is super-dandy, except THEN this thread would need to be moved to the devs forum. By presenting a test that was contested not only by us DSP idiots, but by other developers, it was obvious, even to a layman, that the test itself was at fault. The discussion as pertaining to us DSP idiot musician end users is, afaics, over. If the devs can learn from each other, more power to them. Carry on then, make better products as a result and we'll all be happy. Those who consider themselves half-experts should be allowed into the discussion, as well.

The other issue here (and the one that's produced a lot of amusing give and take) is that some people (myself included) were questioning the AIM behind this thread. If it's a topic for experts, why post it in the instrument forum? To "inform" people? Personally, I still don't see the point: if a synth sounds good to me, do I really need to know WHY? Will it improve my playing? Will it give me a gear hard-on to know that MY synth is more zero delay feedback filter than yours? Will I abstain from buying a synth that sounds good to me if I suddenly discover that it's "not 0df"? Will the "knowledge" gained by doing this "test" make me disregard what my ears are telling me? Did KVR turn into Gearslutz when I wasn't looking? The answer to all of these questions, grasshopper, is "no".

Maybe the posting of this "test" was beset by ulterior motives (which I still believe, and not a thousand shrugs nor holier-than-thou demeanour nor protestations of innocence will change that), or maybe it was just half-baked. No matter how thin you'll slice it, it's still baloney.

And still, somehow, despite all attempts to obfuscate the obvious: Ms.D is the best-sounding analogue emulation on the market right now.