[sdiy] Music

[Scott Nordlund]

> The strings are simple one dimensional Karplus-Strong models. In fact, there aren't even any
> fractional delay filters involved, so very simple.



Yeah, that's kind of like what I did.  But I used 2 delays (should be
very roughly analogous to coupled strings, or maybe prepared piano or
third bridge guitar) and put in some allpass dispersion filters and
tanh nonlinearities.  These are the basic ingredients of "actual"
physical models but I just kind of stuck things together without any
regard for it being correct, or even especially useful, and then
randomized everything.  Surprisingly it still sounded vaguely acoustic.



Results here:



http://www.4shared.com/audio/1lhqZVa5/radiohalo17.html



Worked out better than it should have.




> I've looked into this subject with little success. I've been able to make some fairly convincing
> ride cymbal sounds using FM (total of 16 operators per cymbal though).



That's kind of surprising.  At least as far as the old 2 operator OPL
stuff goes, I always thought the cymbals were the weak point.  But
obviously that was economized as much as possible.


> For a short time, I looked into using some special filters with karplus-strong models, all pass
> with a nonlinearity which moves energy to higher frequencies. I think that many of these are
> needed to adequately model an instrument like a cymbal.



I've read some about modeling spring and plate reverbs.  High order
dispersion filters are very important but it takes a lot of CPU power. 
I'd assume that's true of cymbals and gongs and things too.  Also,
trying to emulate a 2 or 3 dimensional acoustic space with 1D
waveguides isn't ideal.  I think the "proper" way to do it would be ray
tracing, not that it's remotely possible in real time.  There's
probably some easier compromise that gets passable results, though. 
After all, most reverb algorithms make no attempt to actually model an
acoustic process.  They only make a lazy approximation that's fine
tuned to sound reasonable.



> One of these days, I need to play with PD. I keep hearing lots of good things about it's capabilities.



I'm mostly impressed with its workflow, it's certainly one of the
faster ways to get something working, plus it's low level enough that
you don't have to guess a lot about what's going on behind the scenes. 
I've realized it's a waste of time to try to use it the way you'd use a
VST, though.  It's easier to just build up a library of stuff to copy
and paste, then hard wire it together as needed.
 		 	   		  
_________________________________________________________________
Hotmail has tools for the New Busy. Search, chat and e-mail from your inbox.
http://www.windowslive.com/campaign/thenewbusy?ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_1