[sdiy] Timbral musings

[Richard Wentk]

At 13:31 17/02/2003 -0800, harrybissell at prodigy.net wrote:
><snip>
>
>From: Stromeko at compuserve.de
>Subject: Re: [sdiy] Timbral musings
>
>
> >You can take another hint from how electric guitars are played:
> >distortion to my ears often comes to life only when you use it with
> >specific _intervals_, especially containing perfect 4ths (relating to
> >subharmonic scales, 3rd subharmonic), 5ths and 3rds (relating to
> >harmonic scales, 3rd and 5th harmonic).
>
>Not a valid example imho... the problem is because electric guitars are
>polyphonic... and you are driving a polyphonic signal into a waveshaper
>(usually a simple clipper).
>
>The fourths, fifths, octaves sound the best because these intervals
>more closely approximate a single (complex) monophonic waveform.
>
>This is not 'harmonic' distortion anymore (which is pleasant sounding)

Depending on the ratios present, anyway. Lots of odd harmonics will still 
sandpaper the inside of your ears, even though the distortion is nominally 
harmonic.

I've experimented with digital versions of 2nd harmonic distortion and it 
never sounds as nice as it's supposed to. Not on complex signals, anyway.

>its intermodulation distortion (which my dad used to say 'sounds like sh!t')

Well, very transistory, which has its uses...

Some random thoughts:

1. Has anyone tried combining waveshaping with frequency splitting? Shaping 
different frequency bands in different ways might sound interesting. The 
most fun part of it would be that the timbre would become pitch dependent, 
so you could in theory pile on all the distortion at the low end where it 
sounds cool, and keep the top clean.

2. Has anyone experimented with non-linear waveshaping? E.g. an expo could 
make quite a nice shaper. I can see all kinds of interesting problems to do 
with level and DC offset, but I'm sure it wouldn't sound boring. And you'd 
have an interesting degree of voltage control. Of course it would also have 
a tempco, which would either be a bad thing or a good one, depending on how 
you looked at it. :-)

In general I'd imagine there's a lot of space for experimentation by 
deliberately mis-biasing simple transistor circuits.

Or valves. ;-)

3. If you're looking for interesting textures that aren't the standard 
synthy square, saw, etc, it's worth experimenting with granulation. The way 
to do this is to build a really fast AHR envelope generator (say 10us to 
10ms for A & R, maybe up too 100ms for H) controlling a VCA. The AHR would 
be triggered by a VCO, and the output would modulate another VCO via the 
VCA. VC of the envelope coeffs (i.e. A, Hold Time, and R) and the relative 
pitches of the VCOs can create a whole new selection of sounds - great for 
huge cavernous throbs, and unique modulation effects.

Digital granulators are 'polyphonic' and allow the grains to overlap, which 
makes for some fantastic textures and reverb-like effects. But you can go a 
long way with a single mono circuit.

Sorry I don't have schemos - I do most of this stuff with s/ware these 
days, as it's quicker and cheaper than building h/ware - but there's 
nothing too controversial needed here, so it should be easy enough to lash 
together a simple test prototype.

Richard