replies to Jacky

Jacky Ligon write:

>Bill, I've really enjoyed your last two pieces you've posted here

Thanks - praise is always appreciated - (though I'm happy to
receive negative comments as well)

>When you are mapping the timbres with MatLab, are you only changing
>the decay/sustain stage of the timbres, and leaving the attack
>transients and noise components unaltered?

Good question... when I first started doing spectral mappings I tried
several different things. Eventually I settled on a scheme where
I break the sound into two parts, the attack portion and the sustain
portion (I am helped in this division because many samplers have
a "loop" portion which is inherently the sustained part - what comes
before the loop is then, by default, the attack.)
Then I map both the sustain and the attack separately, and
"paste" them back together manually... also, the loop portion
inevitably needs to be re-looped (another manual operation).

>If you are leaving the attack transients unaltered, how many
>milliseconds do you typically leave unchanged?

The attack varies by sound, and is anywhere from a few milliseconds
to maybe 30 or 40 for some sounds. However, I dont leave the
attacks unaltered... the reason for this is that mapping
the spectra of the attacks (as well as the sustained portion)
makes it easier to paste the sound back together... leaving the
attacks unaltered often makes for an awkward (i.e., noticeable)
transition between attack and loop.

>How do you deal with mapping a more complex sound - such as a gong -
>where there may typically be clusters of frequencies close together?

This really depends on what the goal is... for instance,
I have a bunch of cymbal hits that I mapped into harmonic sounds.
The procedure was to identify the largest 20-30 peaks
in the spectrum, and to then map these to the "nearest"
harmonic template. The mapping procedure
was the same as before, but the overall changes were much more
dramatic since the partials moved a lot further, on average.
Maybe I'll post that piece next - for the most part you'd never
suspect that the sounds all began life as cymbal hits.

But this is what I mean by the goal - sometimes, like when doing the
mapping into 19-tet or 10-tet, the goal is to make sounds that
individually remain more-or-less unchanged (as much as possible,
anyway) yet fit together nicely in alternative tuning systems.
When doing more severe mappings, the sound quality will/does
change... most likely mapping things like gongs into a prespecified
system will entail large motion of the paritals and hence
significant changes in the sound...

>I always have loved your use of portamento in your tunes.

Thanks... I'll bet this comes from my "heritage" as a guitar
player... potamento is the synth version of bending strings.

Bill Sethares