[sdiy] Akai S612 filter/schematics

[Theo]

I did the "additive noise sounds" at request for a friend who is producing a
sample cd. The sounds are on his system not on mine, but I'll ask for a copy
next time I go to his studio. Like to have them my self too.

The method used is quick and dirty, never cared about the actual density and
other issues that would need some thought if this had to be a soft synth of
some sort.
What I did was just limiting single cycle random sines to a small
frequency range, write to a long array (about 60 seconds) and over-dub
until it sounded "right". After making some samples for different
"partials",
the samples where mixed to form a static basis sound.

My friend was most interested in sound with some low-end noise and
louder "noise" harmonics in the 1-2 kHz range, so that is what was made.
All the "motion" in the sound was done afterwards with effects.
The "additive noise sounds" them self sounded static,
like a noisy "pieieiep", not very interesting at all.

We also got similar results sending noise thru a software filter that has
the option to draw any frequency response you want, including brickwall
slopes.
This was done in a sample editor, think it was Cool Edit, not sure though.
This method proved less easy to control, but no programming is required :)

Cheers,
Theo



From: Andre Majorel <amajorel at teaser.fr>

> On 2002-04-18 13:36 +0200, Theo wrote:
>
> > > The frequencies of the partials were set by raising a number to
> > > a random power. I think this is a pink distribution (i.e.
> > > statistically, each octave contains as many partials as any
> > > other octave).
> >
> > Ok I see what you mean now.
> > Your energy distribution is right but I think your frequency
> > distribution is different from pink noise.
> > For better sonic result you might try having the partials distributed
> > "linear" and use 1/f for the amplitude.
> > Also the higher the octave the more partials, but keep the energy
> > distribution.
>
> Interesting, I've got to try it. However, without thinking it
> out, I'm afraid it would mean more partials to compute for the
> same subjective noise density, especially in the higher octaves.
>
> Mmm, suppose for instance that you want to generate pink noise
> between 100 Hz and 12.8 kHz (7 octaves), with 1000 partials in
> the band between 1.6 kHz and 3.2 kHz (because that's more or
> less where the frequency resolution of the ear is highest).
>
> With a logarithmic distribution, you need
>
>   7 * 1000 = 7000
>
> partials. With a linear distribution, you need
>
>   (12800-100)/(3200-1600) * 1000 = 7938
>
> partials. OK, just 13% more, but the gap would widen if we
> compared against a weighted logarithmic distribution (where the
> density follows the ear's frequency resolution).
>
> > I've been using similar code to for "additive" synthesis where each
> > "partial" is a small noise band instead of a sine.
> > So far I only can generate static sounds, but sending them thru some
> > effects makes nice atmospheres.
> > Sonically they sit in the spooky category.
>
> I'd like to hear what it sounds like. What band width did you
> use ?
>
> --
> André Majorel <URL:http://www.teaser.fr/~amajorel/>
> std::disclaimer ("Not speaking for my employer");
>