Fw: [sdiy] Control Interfaces (was Wakeman)

[Chris Graham]

>
> What I wonder is if there is some other new way of thinking that can be
> applied to the question of control.  For example, is there a way to use a
> sliding tube arrangement (think trombone or slide whistle) that can
interact
> with a synth as part of the instrument?  Such as an arrangement where an
> impulse (from a speaker?) activates the air column in the tube and some
kind
> of sensor (a microphone?) that senses the impulse and feeds back into a
> circuit that retriggers the impulse to generate a continuous tone much the
> same way a reed interacts with a column of air to produce it's tone.
>

I've done some experimenting with this idea.  The basic premise is that the
usual parts of an all-digital physical modelled wind instrument are a
non-linear driver portion (the mouthpiece) and a linear resonator portion
(the tube).  I got the idea to replace the driver with a real mouthpiece and
couple it to a digital resonator and other digital processing using a
microphone driving an a/d converter, then digital processing, then a d/a
converter feeding an amp driving a speaker.  The mouthpiece would "think"
it's really coupled to a resonant air column.

I built an "instrument" out a small speaker with a cavity behind it made out
of PVC pipe.  The sound from the front of the speaker was channeled  through
a reverse horn to a stem where a mouthpiece could be attached.  The stem
also contained a small microphone pointing at the mouthpiece to pick up the
sound returning from it.  Then I drove the speaker with a feedback loop that
started with the sound from the microphone, fed through a DSP system that
provided a delay, filtering and any other processing I wanted, then to an
amp which drove the speaker.

With a trumpet mouthpiece on the stem pipe, and when the gain, delay and
other conditions were just right, you could play it like a trumpet,
selecting partials with your lip and mouth resonance.  This is just one of a
bunch of configurations I've tried and some worked better than others.  I
think that it could represent a whole new class of instruments in which the
resonant loop is partially analog and partially digital., and I've only
scratched the surface of the possibilities so far.

For example I tried putting a granular processing module in the loop and
produced some really wild sounds.  And it felt much more like playing a real
instrument than twiddling knobs or moving sliders.  You could feel the
vibration of the sound in your mouth/lips like a real acoustic instrument,
and every slight change of your tongue position produced interesting changes
in the sound.

For anyone trying to replicate this, one thing that's essential is to use
VERY low latency adc and dac.  My system had a round trip latency of about
2ms which is really too long.  Ideally it would be less than .5 ms.  I
haven't tried it, but another alternative would be to do the delay as a
analog bucket brigade and use analog filters or other processing, which
would eliminate the adc and dac delay.  One other requirement is that the
speaker must be very powerful and reasonably full range, but have small
diameter.

- Chris Graham