Computer simulation of analog synthesis in general

[John P]

kjarger at jkjelec.com wrote:
> 
> All this talk of the Nord Modular got me thinking:
> 
> Simulation of analog has great promise, IMHO.   Eventually, or even
> now(?), real time emulation of analog circuitry down to the level of
> the Eber's Moll transistor equations, with all of the thermal
> drift, limits (distortion), and 1/f noise factored in, will be possible.
> Imagine scrolling through a list of filter "modules"
> (Mini, SEM, ARP, CAT, EMS, 303...).
> 
> I haven't done any research on the subject (and I'm not looking,
> just wondering), if I were looking to assemble a typical home studio,
> can a typical $2,500 computer
> system and, say $1,000 in extra hardware these days actually,
> simultaneously, run 4 channels of high quality analog
> synth simulation complete with some real time MIDI controller
> response? Without glitches and "slow downs"?

In a word, no.  The $3500 system you describe could probably simulate a 
few modules with high fidelity but not in realtime.  A loaded Capybara 'DSP 
farm' with several 166 MHz processors can simulate a few modules with 
great fidelity and near-realtime response... plus, you get to add $2500 
worth of win95 or Mac.  The total price is over $10K (correct me if I'm 
wrong here..)  I think it'll choke though if you try to simulate a heavily 
patched Moog 55 or multi-panel Serge.   Just too much going on.

Some powerful ANALOG computation goes on in the hardware of your 
typical analog synth.  Nasty differential equations are solved in 
microseconds.  Can't get around that if you want to do a good simulation.

The processors are just beginning to be cheap and powerful enough for us 
to discuss this realistically.  A few more years and it'll probably be good 
enough.  But not yet, not from what I've heard.

John P.