Improvements!

[Urs]

Well, I was thinking the same thing. I like the idea.

[abstractcats]

OHH!! WHat about a granular filter/function/option in Zebra?

[omissis]

Noteworthy: A CS-80 is now awaiting analysis, a Synthex is in reach and a comprehensive collection of Oberheims are ready for dissection. Still waiting for a P5 Rev2 to pop up.

Nice one for the CS! by the way, Urs, let me hint you, the most important component in the CS80 is the VCA...about 10years of experience on the Yamaha CS synths...

If only you would end up modeling the CEM3350 (Chroma)...

[trance_lucent]

pdxindy, Howard.
Well, I use next technique: set Osc to reset phase, volume = 0, volume modulation = envelope. As for envelope, ideally it have to be rectangular and have length equal to one cycle of delay line. So I use v-slope with shape set to most right position, and there is a problem with decay length. I use various length, and with various length it may result in not steady amplitude envelope (if envelope is longer than delay line cycle), or in distortion harmonic content (if envelope is too short, and even if it longer, because second cycle is filled too), and situation is various for the various pitch.

Here are examples.

the first one is for comb prefilled with its own saw:
http://dl.dropbox.com/u/57065234/Comb/saw_prefill.mp3

the several next is for fill with external sawtooth osc:
http://dl.dropbox.com/u/57065234/Comb/o ... _short.mp3
http://dl.dropbox.com/u/57065234/Comb/o ... medium.mp3
http://dl.dropbox.com/u/57065234/Comb/o ... l_long.mp3

and the last is for fill with sine osc (presenting distortion):
http://dl.dropbox.com/u/57065234/Comb/s ... refill.mp3

And, of course, I have to use decay with negative keytracking. May be it should resolve all problems?

[Howard]

pdxindy, Howard.
Well, I use next technique: set Osc to reset phase, volume = 0, volume modulation = envelope. As for envelope, ideally it have to be rectangular and have length equal to one cycle of delay line

Why. i.e. what are you trying to achieve? BTW reducing the oscillator's Key Scale might help. Thanks for the inspiration to get back into the finer side of comb filters!

[trance_lucent]

Why? BTW reducing the oscillator's Key Scale might help. Thanks for the inspiration to get back into the finer side of comb filters!

Well, the main idea is to emulate initial waveform for feeding into Karplus-Strong algorithm. This waveform is determined by the properties of the string and positions of the pickup and exciting position. And it is not random noise like in classic Karplus-Strong. So you need to fill exactly one cycle of delay line with this initial waveform. Keytracking of the osc (and the comb filter) have to be normal and oscillator's pitch have to be equal to comb's pitch. And if you feed delay line longer than one cycle, then waveform in the delayline distort itself, because input signal is added to the signal in the delay line. It can result in not steady envelope, and also in harmonic distortion.

So ideally comb filter have to stop filling itself after the first cycle is filled (so you need square envelope with length equal to one cycle, if you want to emulate this behaviour right now).

And if Zebra will consist this feature with one-cycle-auto-filling, it would be easer to perform controllable synthesis of plucked string. Wavetable plus some noise is very good for prefilling the comb filter.

BTW, second Comb filter with zero keytracking and "dissonant" mode (plus some tuning) is very good for guitar body emulation. Add to this additional eq with some peaks in 100Hz - 1kHz range and you results in very realistic guitar (or another plucked string instrument). Anybody have some docs with Guitar body spectrum response graphs for tuning the comb and eq? I found some, but not very informative.

[Howard]

Thanks for the detailed info, TL. I'd like to try that out too (though - as usual - I'm skeptical about how useful it could be)

[trance_lucent]

Thanks for the detailed info, TL. I'd like to try that out too (though - as usual - I'm skeptical about how useful it could be)

Also please see the last paragraph in my previous message - I decided to edit this post instead of adding another.

And... and - another edit: Such comb filter behavior would be great for Zebrify for "buffer stutter" effect.

[pdxindy]

If you want the square envelop you can use the MSEG...

Also, the env slope parameter can change the decay slope closer to square... close enough I think.

I think it is pretty easy to get a controlled result now... For example, I use 1 osc with a very short env as the comb attack (like you are talking about). Even with the very short envelop, one can modulate it for tonal variety. Then I like to use a second osc with a bit longer envelop and a very low volume to create some string noise...

I also use the second comb for added body like you suggest...

If the comb module could have a parameter like you talk about to control input time, that would likely be convenient for many users who do not figure out the subtleties

[trance_lucent]

Also, the env slope parameter can change the decay slope closer to square... close enough I think.
I also use the second comb for added body like you suggest...

Yes, I use the slope parameter.

And using the second comb for body is very useful. But I noticed that is not so easy to tune this body-comb for convinient low and medium frequencies resonances, because our ears are very sensible to this area, and little changes in this resonance-picture can turn guitar to alt, for example. So adding some resonances by manual eq is very useful. Is there in internet some good measurements of guitar (violin, alt, and so on) body response?

[Howard]

Is there in internet some good measurements of guitar (violin, alt, and so on) body response?

Yes please - anyone?

[hakey]

Found various academic papers in the past:

http://www.phys.unsw.edu.au/jw/violintro.html

http://www.laguitarra-blog.com/wp-conte ... guitar.pdf

http://www.wseas.us/e-library/conferenc ... AMTA04.pdf

...but then struggled to translate them into, say, a zebra patch in any meaningful way.

I mean, Zebra's EQ isn't flexible enough to replicate the subtleties you see in these freq response graphs. I suppose you could use both EQ's and various peaking and shelving filters, but even then what's the process for accurately translating a freq response graph into Zebra in this way?

[Cremisi]

Noteworthy: A CS-80 is now awaiting analysis, a Synthex is in reach and a comprehensive collection of Oberheims are ready for dissection. Still waiting for a P5 Rev2 to pop up.

I think I could almost cry

[trance_lucent]

Found various academic papers in the past:
http://www.phys.unsw.edu.au/jw/violintro.html
http://www.laguitarra-blog.com/wp-conte ... guitar.pdf
http://www.wseas.us/e-library/conferenc ... AMTA04.pdf
...but then struggled to translate them into, say, a zebra patch in any meaningful way.
I mean, Zebra's EQ isn't flexible enough to replicate the subtleties you see in these freq response graphs. I suppose you could use both EQ's and various peaking and shelving filters, but even then what's the process for accurately translating a freq response graph into Zebra in this way?

Thank you, I'll check this.
Usually I use external Logic's EQ's, sometimes several in series. But sometimes use Zebra - there are 4 peaks in two EQs, and this is often enough for main resonances representing crude shape of the instrument's body. It is only need to add additional resonance picture with Dissonant Comb, using Keytracking = 0%, tuning Tune, Tone and Flavour knobs. Also with luck sometimes it is possible to manage without EQ resonances and use only that comb filter, but it is need careful tweaking of Tune, Tone and Flavour knobs. And great luck again.

[hakey]

Also with luck sometimes it is possible to manage without EQ resonances and use only that comb filter, but it is need careful tweaking of Tune, Tone and Flavour knobs.

My attempt at a classical guitar in Diva - no outboard processing (mainly a lot of tweaking/comparing against the real thing, and some luck ):

Classical Diva