[sdiy] ... Simulating a Moog

[Richard Wentk]

At 10:13 03/05/2004 -0700, Julian Bunn wrote:

>I guess I'm agnostic on this point.

That's probably wise.

The current state of play is that analogue synths still sound more like 
analogue synths - or at least *good* analogue synths sound more like good 
analogue synths - than any softsynths do.

But softsynths win easily on reliability, polyphony, versatility, cost, 
size, and ease of transportation. And the only musical application for an 
old analogue is to sound like an old analogue. You can do a lot more with a 
good digital synth, especially patchable designs like Reaktor and Max/MSP.

Since my main interest is musical and not technical - I get much less of a 
kick out of building things or solving problems than I do out of trying to 
make interesting music - I'm finding my interest in hardware is waning.  If 
I want a step sequencer, or a 10-oscillator uber-FM synth, a custom vocoder 
with some quirky features, or an additive synth, it's a lot easier and 
quicker to patch one together in Reaktor than it is to build one. (Like - 
an evening or two vs months.) I then have that instrument available on disk 
to order whenever I want to use it again - which is a lot more convenient 
than spending an hour or so breaking down one patch and building another.

Controllability really isn't any kind of an issue. There are plenty of 
knob-laden external interfaces now. In fact you can get more control from 
them than from a modular because everything is one place instead of spread 
out over square feet of panel space, and you never lose knobs or indicators 
behind patch cables.

The one downside is that today's screens are still too small. A laptop with 
a 21" screen would be a very wonderful thing. A 12,000 by 8,000 pixel wall 
projector would be even better.

> > Good call! This, in fact, is the approach that Korg claims to
> > have used for
> > their new (now shipping?) Legacy collection. Korg calls its technique
> > Component Modeling Technology (CMT). More info here, scroll
> > down the page a
> > bit:
> > http://www.korg.com/gear/info.asp?A_PROD_NO=Legacy
> > As for the results, indications are that Korg have done quite
> > a good job.
>
>Oh yes, this is exactly what I was imagining! Thanks for the pointer.

As others have pointed out, the component approach is a good one but it's 
also beyond the reach of today's DSP horsepower. It's not just a question 
of modelling parasitic connections, but also of the quality and accuracy of 
the models used. Real capacitors are only notionally related to virtual 
ones, and there's a lot that a real capacitor does that a simple charge 
tank model won't cover. You can certainly build more complex models that 
more closely match what happens in the real world, but it's a time 
consuming and rather specialised job.

But then so is soldering clones of old hardware together, so the argument 
works both ways. :-)

I think many of today's VA's and softsynths already sound as musical - if 
not a lot more musical - than cheap and rather forgettable analogue designs 
like the Jen SX1000 and the Transcendent 2000, but are some way behind 
classics like from PPG, Moog, ARP and the rest.

But I think there will come a time, certainly within ten years and possibly 
within five where there will be no reason at all not to use the simulation 
approach. With enough bit resolution and a high enough sample rate, and 
careful enough modelling, I don't see any reason why digital techniques 
won't be able to accurately simulate anything and everything that's ever 
been done using hardware.

But I think the pseudo-Spice approach is more likely to do this 
successfully than trying to characterise transfer functions. Analogue 
hardware is more complex than it looks, and DSP techniques that are 
transfer-function based absolutely fail to capture a lot of the subtleties, 
which may change according to DC offset (not rare in a modular synth), 
temperature (including self-heating) and other semi-random processes.

> > For what it's worth, there's one thing about the real-vs-emulations
> > arguments that bugs me: You'll find people saying they sound
> > different, the
> > implication being that the emulation is inadequate. Yet, they
> > will also say
> > that no two examples of an [insert vintage synth name here]
> > sound alike,
> > either. (However, some softsynths deserve the criticism.)
> > --
> > john
>
>This reminds me of discussions about speaker cable ... I think some
>people just have very discriminating cochleas, or vivid imaginations :-)

I know some people who used to run a company stripping wiring out of old 
military aircraft and repackaging it as super-exotic worth-more-than-gold 
speaker cable.

I've also heard at least one person say that they only like listening to 
their hifi in the early hours of the morning because the mains supply is so 
much cleaner then.  :-)

Richard