CW VA synths behave weird (noise, polyphonic mode)

darkrezin · Post by **darkrezin** » Wed Jan 04, 2006 4:40 pm

I agree with some of your points Kybernaut. I would like to see a greater variety in 'reaction' of the synths with stuff like envelopes etc. Two things I'd like to say though:

1. IMHO this kind of stuff affects most, if not all, digital synths

2. I think it's good to take instruments for what they are and what they can do, rather than what they are not. For this reason I personally really hate it when emulations are marketed.

IMHO there has never been, nor will there ever be, a true emulation of analog. At the very least it will require 384khz sample rates and next-generation algorithms using stupidly huge amounts of DSP power. Not to mention being able to model every single aspect of how an analog synth musically responds.

If you listen to some synth-heavy jazz and funk from the 70s for example, you'll realise there are many many many ways of getting infinite sounds and variations out of analog gear. To model everything is not easy. Any of the current digital emulations do not come close at all.

Post by **hubird** » Wed Jan 04, 2006 6:16 pm

Any of the current digital emulations do not come close at all
in this context it is also interesting that serious tests taken with serious analog gear specialists show that even them were hardly capable of guessing what sound was made by which one of the used and named analog synths involved in the tests...

kybernaut_01 · Post by **kybernaut_01** » Wed Jan 04, 2006 6:37 pm

On 2006-01-04 16:40, darkrezin wrote:
1. IMHO this kind of stuff affects most, if not all, digital synths

2. I think it's good to take instruments for what they are and what they can do, rather than what they are not.

I'm absoluteley with you. And believe me, I only take the time to care about this stuff because I like CW synths so much. It's not that I want to bash their synths here. They are way ahead of anything else that has been made in the digital VA realm.

kybernaut

darkrezin · Post by **darkrezin** » Thu Jan 05, 2006 11:33 am

hubird - it depends what they were doing in the tests. It might not be so apparent when playing certain types of generic patches but say if you do some heavy FM on oscs it becomes much more apparent. Because most people use synths in a pretty generic way it doesn't matter to most people but if you like what happens to analog gear when it's abused with weird patching and taken to its limits then it's not really possible to get the same results with digital emulations yet.

Everything depends upon what you're doing and how you want to use the technology. But to say that it's a true emulation is incorrect.

[ This Message was edited by: darkrezin on 2006-01-05 11:34 ]

Post by **hubird** » Thu Jan 05, 2006 5:03 pm

you're right, I really think so

In fact I was adding to the 'relativity' argument.
If even specialist by ear can't really distinguish between (analog) synths, why bother so much about if the emulations are perfect

But that's me, if an emulation sounds good, it's perfect to me

Shroomz~> · Post by **Shroomz~>** » Fri Jan 06, 2006 2:25 am

On 2006-01-03 16:53, johnbowen wrote:
the polyphonic/monophonic noise issue is something the designer of the synth can set. For some of my synths, I put a switch in to choose, but usually I make it a mono noise signal, since I personally prefer not to have the 'comb filter' effect when I play polyphonically.

Is this 'comb filtering' deliberate? Is it implemented or unwanted residue/side FX? I was wondering if it's there to diguise or attempt to smooth out 'noise aliasing' when playing digital OSCs polyphonically.

kybernaut_01 · Post by **kybernaut_01** » Fri Jan 06, 2006 4:58 am

On 2006-01-06 02:25, Shroomz wrote:

Is this 'comb filtering' deliberate? Is it implemented or unwanted residue/side FX? I was wondering if it's there to diguise or attempt to smooth out 'noise aliasing' when playing digital OSCs polyphonically.

I think I should clarify the term. In this case, it's certainly not a "filter" implemented by Creamware with intention, but rather a side effect.

In signal processing, a comb filter adds a slightly delayed version of a signal to itself, causing phase cancellations. The frequency response of a comb filter consists of a series of regularly-spaced spikes, so that it looks like a comb.

Example Graphs

How many of these spikes occur and how close they are to each other depends on the amount of delay that is added to the original signal.

Since the spikes change the frequency response of the original signal (in this case white noise) it is no longer white noise and hence it sounds different.

Comb filters can also be used for creative sound design and are featured in some synthesizers (ie. the Waldorf Q) and are usually implemented with modulatable delay and feedback controls. They are also vital for certain approaches to physical modeling ("Karplus Strong").

Back the topic of this thread:

In an analog synthesizer the noise source can be monophonic or polyphonic. In the first case the output of one noise generator is fed proportionally into the different voice chains. In the latter, each voice chain has its own noise generator. The levels add up when sounds are played polyphonically. Even in this second case, there will be no overlay and "comb filtering" effects because each noise generator outputs its own (completely random!) signal.

Creamware seems to use one single noise generator but seems to create several copies of it for each polyphonic voices. Since these are no longer true "random" signals they are related to the original source. For a reason that I don't know there seems to be a delay in the internal processing.

cheers,

kybernaut_01

[ This Message was edited by: kybernaut_01 on 2006-01-06 04:59 ]

Shroomz~> · Post by **Shroomz~>** » Fri Jan 06, 2006 5:15 am

Maybe just a little slip of sloppiness. Nothing when compared to the fact that Clavia's G2 modular is programmed to calculate it's delays in 16-bit... now that's sloppy, but then the G2 is a deep synth & must have needed optimised to such an extent.

You'll just have to play CW Scope Synths mainly in mono & use John Bowen's for smoother poly work

astroman · Post by **astroman** » Fri Jan 06, 2006 5:19 am

On 2006-01-06 04:58, kybernaut_01 wrote:
...For a reason that I don't know there seems to be a delay in the internal processing.

well, (my memory is not always reliable) but as mentioned earlier when I tried to verify your examples, I got a similiar result and then remembered the FX page might be active (I choose an arbitrary preset) and could provide the delay.
When switched off explicetely the effect at least significantly decreased (if not vanished at all).

The FX unit might not be taken (completely) from the signal path even though the screen indicates it is. Imho the NOAH is even more suspective in this context as it's more or less remote controlled.

cheers, tom

[ This Message was edited by: astroman on 2006-01-06 05:20 ]

kybernaut_01 · Post by **kybernaut_01** » Fri Jan 06, 2006 5:46 am

On 2006-01-06 05:19, astroman wrote:

well, (my memory is not always reliable) but as mentioned earlier when I tried to verify your examples, I got a similiar result and then remembered the FX page might be active (I choose an arbitrary preset) and could provide the delay.
When switched off explicetely the effect at least significantly decreased (if not vanished at all).

The FX unit might not be taken (completely) from the signal path even though the screen indicates it is. Imho the NOAH is even more suspective in this context as it's more or less remote controlled.

My examples have been done by turning up the noise level in the mixer section. The filter is completely open. Envelope amount, resonance, keyboard tracking and all modulation amounts are set to "zero". Then, I played some 5 notes on the keyboard (kept holding down the notes that I pressed first).

Would be interesting, to hear the results of other incarnations of Creamware's Prophet emulation, such as Profit-5 (Scope), Pro-12 (Scope), or Pro-12 (ASB).

Regarding your observations with the FX unit, I was not really able to reproduce that on my NOAH (as you probably know, NOAH synths do not have a built-in FX section, all FX are centrally managed). Until now, I never had any problems to completely turn FX off. Also, if the comb filtering issues I describe are really dependent on the FX unit (as you assume), then it would apply to all signal sources of the synth (all oscillators) and not just the noise generator.

Again, I'm not sure what CW did but I know the results: Cancellations in the frequency response.

cheers,

kybernaut_01

astroman · Post by **astroman** » Fri Jan 06, 2006 8:23 am

I really have no idea about the internal structure of the Profit, let alone NOAH - so this is just speculative.
The noise may even have a different path from the oscillators as I'd guess there's only one source anway and not 6 truely independant random generators (Shayne may know more in this context)
I could swear that when I did what you describe above (just 'keying' the noise with all modulations decoupled) it was just noise adding up and down.
But (now) I do suspect that there are 'random' changes in my setup that only reveal on concentrated listening level on very detailed sources.

cheers, tom