[sdiy] Analysis of frequency variation in analogue synths

[Colin f]

> > Clearly things will be different with a two oscillator sound.
> > A worthwhile test there would be to see if you can pick out a pair  
> > of real
> > analogue oscillators in unison, against a pair of samples 
> in unison  
> > where
> > the sampled oscillators have small amounts of random modulation.
> 
> That would depend on the kind of modulation.

> What makes it complicated is that each item on the list has enough  
> potential detail to spawn a few PhDs.

My point is that there is an untested assumption here about the degree to
which it depends on the kind of modulation.
Before you go to the effort of carefully analysing all the possible
influences, it's worth testing empirically whether a fairly simple
combination of low and high speed cyclic variation plus a random component
modulating a digital oscillator's pitch and amplitude sounds any more or
less analogue than a real analogue VCO.
Reproducing detail that you can't hear is wasted effort.

> And there will be no digital:
> 
> Splashy phase shifts at the top end
> Quantisation noise
> Aliasing
> ...and other nasties that contribute to that crunchy grainy digital  
> sound that seems almost inevitable when working at 44.1.

If a minimoog still sounds good on a CD, then the medium isn't the problem.
In terms of actual modelling of the sound, clearly you need to have much
greater bandwidth and bit depth in your DSP than the final output format to
take in the full spectrum of modulation products possible with a smooth
top-end roll off, otherwise you are going to get artefacts.
But that's just a question of throwing enough CPU power at it.
The failure of analogue modelling so far to fool the hard-core of analogue
fans is probably more economic than technical.
Digital photos 10 years ago looked crap, but now they rival the best 35mm
film cameras.
There's nothing spectacularly new being done - just increased resolution.

> As an aside - it's a classic case of reality being much more  
> complicated than the usual models. *Accurately* modelling analogue  
> behaviour is light years away from putting a digital sawtooth 
> through  
> a digital SVF and using digital envelopes and LFOs for modulation.

I'd be amazed if anyone could pick out a modern digital envelope generator
from a hardware analogue EG in a blind test, with all other things equal,
i.e. CPU driving DAC driving VCA versus EG chip driving VCA.

Cheers,
Colin f