[sdiy] stringfilter

[JH.]

> why 20 bands per channel? 

40 Bands across the whole audio range.
I'd say this reds as "a lot of bands", and 40 is about what you
can do easily. A real string instrument has a lot more.
Apparently 40 are good enough to get some impression 
of sounding similar.

>assume each of the band is a bandpass filter with 
>its own resonance frequency. But I'm only aware 
>of the 3 main resonances for each string 
>instrument.

The 40 resonance peaks are in _addition_ to the 3 main resonances.

The idea was modeling the 3 main reonances, plus add some
overall comb filtering (40 resonances) for further enhancement.

>what else would 
>be necessary to get the "stringy" character that 
>Juergen demonstrates on his webpage?

A fairly reasonable input (saw or pwm, with a string-like
amplitude envelope), and a good reverb.

>And I agree very much, it would be *so* 
>interesting so hear Ken Elhardt giving us a few 
>hints how he generated his string sounds!!

I'm sure he'll answer this himself if he wants. 
He told me once, and without telling any secrets I can say that he
used rather unsuspecting commercially available FX boxes -
and his exceptional talent to hear details in sound to set
all the parameters right. 

JH.