KVR Audio

braj · Post by **braj** » Mon May 21, 2012 2:03 am

'The motion of the ocean' defense eh?

olikana · Post by **olikana** » Mon May 21, 2012 6:40 am

lotuzia i hope u realize now that this

lotuzia wrote:A commonly held belief is that non-0df filters are
unable to produce sounds with rapid onset transients, like
percussion because of the delay introduced. This is incorrect,
the 1 delay lag is not perceived by the ear. The 0df filter
problem affects sounds at more or less mid resonance settings, when
the filter frequency is varying (when is generally the case).

should have been this instead

a commonly held belief is that non-0dfF filters are unable to
accurately produce analog sounds with fast modulations (expample:percussions,bass and audio rates) because of the delay introduced. this is correct and it is audible with mid resonance settings on 24db ladders/diodes. in this case when fast modulated the 1 sample delay lag (or the conequences of it) can be perceived by the ear .

and where did u get this "rapid onset transients" term from is a
mystery or simply u misunderstood the issues which musicians were
bringing to developers, cos as far as i remember(since the first thread about 0 delay feedback filters opened by me almost 1 year ago) it was always about fast modulation .

and it's even more a mystery why you and xavier (although u partecipated both in tha oldt thread) did not mention at the time that the xils synths had 0dff (the thread specifically asked if any synth out there had 0dff) nor you had anything to object to Urs saying in that thread there was no native solution at the time, if your synths had 0ddf since 3 years?

i personally don't care....soon all filters will be 0dff (filters without 0dff sound simply wrong- the 1 sample delay was only ever there to make up for dsp deficiency).....but unrecognizing other developers achievements is lame.

Lotuzia · Post by **Lotuzia** » Mon May 21, 2012 8:44 am

olikana wrote:lotuzia i hope u realize now that this

lotuzia wrote:A commonly held belief is that non-0df filters are
unable to produce sounds with rapid onset transients, like
percussion because of the delay introduced. This is incorrect,
the 1 delay lag is not perceived by the ear. The 0df filter
problem affects sounds at more or less mid resonance settings, when
the filter frequency is varying (when is generally the case).
should have been this instead

a commonly held belief is that non-0dfF filters are unable to
accurately produce analog sounds with fast modulations (expample:percussions,bass and audio rates) because of the delay introduced. this is correct and it is audible with mid resonance settings on 24db ladders/diodes. in this case when fast modulated the 1 sample delay lag (or the conequences of it) can be perceived by the ear .

and where did u get this "rapid onset transients" term from is a
mystery or simply u misunderstood the issues which musicians were
bringing to developers, cos as far as i remember(since the first thread about 0 delay feedback filters opened by me almost 1 year ago) it was always about fast modulation .

and it's even more a mystery why you and xavier (although u partecipated both in tha oldt thread) did not mention at the time that the xils synths had 0dff (the thread specifically asked if any synth out there had 0dff) nor you had anything to object to Urs saying in that thread there was no native solution at the time, if your synths had 0ddf since 3 years?

i personally don't care....soon all filters will be 0dff (filters without 0dff sound simply wrong- the 1 sample delay was only ever there to make up for dsp deficiency).....but unrecognizing other developers achievements is lame.

Hi Olikana,

Yes, opposite to you, we specified that non 0df filters can sound good, even if Xils are not of this type. Thats what we think. Then we gave some explanations so that people can understand the real meaning of 0df filters.

Actually we think a synthesizer is more than a pile of technical specs.

This is in the exact same spirit that we participated in this thread you mentioned in the last 2 pages, to discuss oscillators, and not the main topic. So that I was at that time on hollydays, we were mainly concentrated on the Synthix release, and I did not even read wht Urs or other people wrote at the beginning of the thread.

As for "unrecognising other developers achievments" : yes you're right it might be lame, but it was not our case

We never said that a synth pretending to have 0df filters had not for example. Thats why we were very suprised that some people declared back in Nov 2011 that the Sytnhix did not have 0df filters.( not only without any proof for that, for good reasons, but also probably whithout any understanding of the in and hows of 0df filters )

And btw is this not very similar to what you did yourself, when in a recent thread you emitted a very patronising comment on Xils synths, when I just said thay had 0df filters ?( Answering me Something like "...... I let you to your ridiculous zaps" iirc etc )

But opposite to you I wont try to give you moral lessons etc. You're free to dislike Xils synths, and to make it show in a more or less subtle way. Et voilà

Lotuzia · Post by **Lotuzia** » Mon May 21, 2012 8:54 am

Urs wrote:
Lotuzia wrote:Xils did not target Diva . Never.
That might be true.

But then, you have become the public face of XILS, not least since you organized that long Synthix announcement thread last year. And this is why people associate whatever you say with XILS, and it doesn't matter whether it's posted on your own merrits, or on your sound shop or on XILS. Thus, whatever you say is commonly perceived as a XILS opinion.

Yes no problems with that.

Perception is not equivalent to truth tough. it can reflect the same reality, as it can be eventually a source of errors.

Now if people perceive me beeing sad for Donna summer passing as a Xils opinion, or that the last soundbank we released for fm8 is also a Xils opinion, then what can i do ?

Anyway all this is shown in my sig for a long time now and therefore there no ambiguity about all that.

Urs · Post by **Urs** » Mon May 21, 2012 9:09 am

Lotuzia wrote:Thats why we were very suprised that some people declared back in Nov 2011 that the Sytnhix did not have 0df filters.( not only without any proof for that, for good reasons, but also probably whithout any understanding of the in and hows of 0df filters )

I can look that quote up. I'm pretty sure I used the word "sceptical" rather than "certain". That means, I left it open even though I had (and still have) doubts.

Unlike back then I'm now also certain that the Synthix filter is indeed oversampled, but the downsampling filter isn't as steep as I'm used to.

We all make mistakes, don't we?

(but thanks for doubting my understanding of 0-delay-feedback filters after giving you half the source code to implement them)

mkdr · Post by **mkdr** » Mon May 21, 2012 9:28 am

Who invented the zero delay feedback filters?
Is the data of implementing them publicized somewhere?

Lotuzia · Post by **Lotuzia** » Mon May 21, 2012 10:18 am

Urs wrote:
Lotuzia wrote:Thats why we were very suprised that some people declared back in Nov 2011 that the Sytnhix did not have 0df filters.( not only without any proof for that, for good reasons, but also probably whithout any understanding of the in and hows of 0df filters )
I can look that quote up. I'm pretty sure I used the word "sceptical" rather than "certain". That means, I left it open even though I had (and still have) doubts.

Unlike back then I'm now also certain that the Synthix filter is indeed oversampled, but the downsampling filter isn't as steep as I'm used to.

We all make mistakes, don't we?

(but thanks for doubting my understanding of 0-delay-feedback filters after giving you half the source code to implement them)

Cool down Urs, please, I was refering to Hakey, not to you !!! ( and was delicate enough NEVER to quote you in this thread on this particular subject)

So this therefore implies that I'm not questioning at all your understanding of the implementation of 0df filters !. This would be completely silly/stupid from me, as, and I say it clearly in this thread, it would be a FALSE statement from me. But I never said that so its ok.

Its however false that you gave us, like Xavier made you the remark in another rather old thread, half of the code to implement it : Xils is working on the 1df problem for 3 years now, with 2 generations of algorythms !! So I'm sorry but you just can't say that : Our solutions owe nothing to you, as your solution owe nothing to us. We have very CLEARLY stated from the beginning that several solutions are indeed possible to solve the 1df problem. Mystran proposes another one btw.

And yes we all make mistakes. That's true. No worries.

More than that, and trying to go a bit deeper : I just wrote in another thread that we welcomed you in this thread. This remains true. You agreed we never targeted Diva, and this is true. Actually we never targeted any particular synth, we merely think that synths have their own identity.

OT : obvious but I prefer to state it clearly : Its of course NOT our responsability that, during this particular thread, you were attacked personnaly by some people. it could have happened in any other thread (and btw it just happened in some other threads as well) Its quite obvious that we have to face some common problems here, as we are also attacked very often. So I know that you dont think that we're responsible but well maybe its better to state thios clearly.

I'd just want to add (on a personal side) that when some people attack you in a very aggressive way, then its obvious for readers who is wrong. The more subtle ususal passive/aggressive thing will fool people for a while, but repeated and systematic attacks will show, so if I were you, I would not worry too much about all this. Life is much more that an interwebz battle on a forum, and people whose only aim is currently continuously bashing and trashing other people's work will eventually find a real life and better reperesentation of themselves, and then vanish as suddenly as they appeared, because obviously music is not their first concern. As for we, we have to build nice musical instruments for musicians, and this is such a nice work that everything else should really be seen as minimal collateral damages and infinitesimal pollutions.

Urs · Post by **Urs** » Mon May 21, 2012 10:21 am

You are right. I worry too much.

How about we put the grudges aside and file them under experience?

Richard_Synapse · Post by **Richard_Synapse** » Mon May 21, 2012 10:33 am

Cyforce wrote:How big is the quality difference between a "normal" filter, just for example basic ladder 4-pole lowpass, compared to one which is also a ladder lowpass but with zero feedback delay?
Logical on the same samplerate & oversampling settings

Is this difference really so big? If the difference would be only small, the question "would it be worth it, through the higher cou usage?" would very vaild.

Good question- this depends on how the non-0df ladder is implemented. If it's implemented well, it can get close, while saving a lot of CPU in the process. The 0df filters are easier to work with, though. From my experiments they generally seem to have less problems, e.g. at high cutoff or resonance settings, using extreme modulations, etc etc.

In Dune 2 we might offer both types, as there's quite a few sounds where 0df filters are just major CPU spew. They may even degrade the sound quality for patches where a clean sound is desired.

Richard

carrieres · Post by **carrieres** » Mon May 21, 2012 10:35 am

Urs wrote:You are right. I worry too much.

How about we put the grudges aside and file them under experience?

Thank you Urs, i was depressed that even dev need to attack other dev
the current mood remind of j!development, the dev just disappeared from the scene !
i hope no other dev will be that discouraged to give up on synth development

carrieres · Post by **carrieres** » Mon May 21, 2012 10:41 am

Rich, you are still there focusing on what really count : the sound (with cpu cost in mind).
i dream that you would have several arms to code all your synth hidden in your brain, what about Shiva as your next synth name

Cyforce · Post by **Cyforce** » Mon May 21, 2012 10:50 am

Richard_Synapse wrote:
Cyforce wrote:How big is the quality difference between a "normal" filter, just for example basic ladder 4-pole lowpass, compared to one which is also a ladder lowpass but with zero feedback delay?
Logical on the same samplerate & oversampling settings

Is this difference really so big? If the difference would be only small, the question "would it be worth it, through the higher cou usage?" would very vaild.
Good question- this depends on how the non-0df ladder is implemented. If it's implemented well, it can get close, while saving a lot of CPU in the process. The 0df filters are easier to work with, though. From my experiments they generally seem to have less problems, e.g. at high cutoff or resonance settings, using extreme modulations, etc etc.

In Dune 2 we might offer both types, as there's quite a few sounds where 0df filters are just major CPU spew.

Richard

My major concern is on the hole zero feedback delay issues, which takes here at KVR a lot of pages and small fights - is it really worth all this trouble?

Sure without a doubt, all user's would love to have filter's which sounding just better as on previously synths. But see this...

Some years ago, where the reference cpu was Pentium 4, the current synths in this time period could be used without any problem up to 10 instances or more.
Than came the dual core and all people were happy "oh wow 70% more performance" etc.
As musician, logical, you need performance because a track is not compost with only 1 synthesizer and 1 effect, a lot more.
Today, we have cpu's like i7 with a huge performance potential. But we can use the synths of today(like D-CAM, Razor, ACE, DIVA etc) never with up to 10 instances, also on normal quality settings.

So i think, maybe the run for the maximum of quality - or better said the run for expand the limit of possible maximum quality - goes just to fast.
In earlier times, synth's and also their filter didn't sounding bad. And today sure they sound better but for much cpu and with up to 16x oversampling, where years ago thinked "well 4x, looks quite overdoesed"

And the filter is only one part of the hole engine of the synth, also if it's the main "middle" module, where the osc sound goes in and going forward to the end effects.

Maybe it would be better instead of trying to expand the quality limit more and more and more with new ideas/codes/methods etc, to expand the effectivity?
What we have from synths which sounding superb, but we couldn't use them as we would like to?....

/edit

carrieres wrote:Rich, you are still there focusing on what realy count : the sound (with cpu cost in mind).

That's in my mind probably the best way, or the way which should be... +1 for Synapse Audio

george · Post by **george** » Mon May 21, 2012 2:39 pm

Gentleman, now Corona features a zero-delay feedback filter. You can enable it under Filter > Mode button at the bottom right (Normal or ALT <- zero delay feedback)

Hope you like it (they are amazing)

Cyforce · Post by **Cyforce** » Mon May 21, 2012 2:48 pm

george wrote:Gentleman, now Corona features a zero-delay feedback filter. You can enable it under Filter > Mode button at the bottom right (Normal or ALT <- zero delay feedback)

Hope you like it (they are amazing)

Great, specially that both variants are onboard(atl + zfd), so a great possibility to compare both types and listen to the difference.

george · Post by **george** » Mon May 21, 2012 2:50 pm

I'm in love with the zdf ones

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