Hi, there

If this is off topic, please let me know.


Since I first got intersted in csound and dsp, I have always been
wondering whether there are some methods when I want to generate a
specific sound?  As a programmer, I know that there are a lot of
algorithms to realize some result I expect.  But as for sound
generation, I have learned AM, FM, waveshaping... but have no idea
about what to do when I really think out of a sound.  What I have to
do is just trying again and again.  Sometimes I got what I want, but
it is very seldom for me.  Could someone tell me what to do?

Regards




==
Qian Chen

_________________________________________________________
DO YOU YAHOO!?
Get your free @yahoo.com address at http://mail.yahoo.com

At 4:09 AM -0800 2/2/98, Qian Chen wrote:
>Hi, there
>
>Since I first got intersted in csound and dsp, I have always been
>wondering whether there are some methods when I want to generate a
>specific sound?

Inverse problems are often very difficult. Csound and other related
technologies are very good at answering the question: If I use this
algorithm, what will it sound like? The inverse problem: If I want this
sound, what algorithm do I use?, has an infinite number of solutions, and
an infinitely larger number of non-solutions. Those who are best at it tend
to use a combination of intuition, analysis, and a trained ear.

My personal heuristics are:

1) If I want the sound of a physical instrument (e.g., a violin) I hunt
down someone who knows how to play one, and with liberal quantities of food
and beverages (and promises of fame and glory) try to persuade them to play
it for me.

2) If I want lots of close-to-physical instruments, I use samples of the
same. I'll occasionally use hetero/adsyn, lpanal/lpread, or pvanal/pvoc on
the samples to get what I want.

3) If I want synthy-electronicky sounds I rely on my 25+ years of
experience with oscillators, signal generators, and analog synths, and my
15+ years of experienc with digital synthesis to pick a starting point, and
then I tinker until it's close enough.

4) I play with all of the new opcodes in Csound when I have the time (I'm
presently way behind).

5) When someone poses an interesting question on the list I spend a little
time looking for a solution or two.

If you know enough about your sound to describe it EXACTLY in terms of
frequency content versus time, there are a number of methods in Csound to
realize it exactly. Unfortunately, people usually think of sounds like: I
want a cross between a piccolo and a tuba that plucks and pitch-bends like
a guitar, but I want it to sing Swahili and sound like it's being played in
the National Cathedral. All of the experienced Csounders already have their
minds going on how to do it, but there is NO general algorithm to get
there. Sorry.

-------------------------------------------------------------------------------
Erik Spjut (rhymes with cute) - Acting Director,The Center for Design Education
and/or Associate Professor of Engineering
Harvey Mudd College, Claremont, CA 91711  USA
Erik_Spjut@hmc.edu      Ph & Voice mail (909) 607-3890      Fax (909) 621-8967