14 Apr
2013
14 Apr
'13
11:05 a.m.
[Fons Adriaensen]
On Sat, Apr 13, 2013 at 06:12:42PM -0400, Paul Davis
wrote:
The problem with the additive approach is that it doesn't easily model
nonlinearities and the repeated excitation of an already oscillating
instrument properly. It's also hard to correctly emulate the
behaviour of continuously varying excitation points. I suppose you
could go a long way with precomputed data but the multiple dimensions
of the parameter space, with time being one, might make this a bit
cumbersome.
I'm not qualified to judge the realism but to me this snippet sounds
very good in an absolute way.
Tim
.. the usual "mode" operators in Pd,
CSound, SC etc, don't really
capture the subtleties required.
Modelling it is one way, and that could indeed become quite
complicated. Another way is to analyse the real thing, find out the
parameter space, and recreate the sound using additive synthesis, no
matter what could be the model behind it. Sort of cheating, but it
works.