Hi,

I usually try to keep the kr near 1000, because that's nearly the same rate
that MIDI is. (Yes, I know that MIDI is not optimal, but it's there and
even Csound can make use of it.) So the values I use are:
44100, 1050, 42
22050, 1050, 21
11025, 1225,  9

(perhaps its better to keep the kr=1225 for all three versions and only
change the sr and ksmps?)

I think it really depends on the machine and problem/piece/instrument which
default is the best. It would probably be a good idea to add the default to
the Csound preferences, so the user can select his own standard. For a
fixed one I would prefer sr=22050 and kr=1225 or 1050. I don't thnk it is
the right time to set the standard to 44100, most of the machines are still
to slow to play a piece of music in real time. I would prefer the real time
output over sound quality when working on instruments or music over high
quality output. hen I think I need a better quality to decide how to go on,
I will compile a version out of real time with sr=44100.

About that GM stuff...:
I don't see any reason to discuss this here. The intention of Csound is not
the intention of GM. It is no problem to make samples of several
instruments and write an instrument to play them (I think with loscil).
Csound is not designed to handle all kind of MIDI messages (i.e. program
change is difficult to handle), so I wouldn't expect to see a clean
solution. I'm sure a lot of globals are needed to handle pgm change and
loading of sample tables. GM in Csound needs a really fast machine if you
want more than just a few voices and I'm sure most of the GM pro's are not
satisfied with a sample rate of 22050 or below.
I use MIDI for Csound to control my instruments in real time (with some
delay, of course), but if there would be something faster and more
intelligent designed like ZIPI, that I can use I would change to it as fast
as I could. Unfortunately I don't know of any device using ZIPI (there ay
be some prototypes at CNMAT?) and Csound also doesn't know ZIPI, so I have
to stay with my MIDI gear. Sometime I have to build the MIDI controllers
myself because they don't exist or the implementation is not adequate to my
needs.

For all of you waiting for my physical modeling bibliography:
It is still delayed because the beginning of my thesis is delayed. I hope
I'm finished during April (rather the end than the beginning) with the list
(400 articles are waiting to be typed in). At the moment I can advise you
the books by Benade and Fletcher/Rossing, the two or three issues of CMJ
and the homepage of Julius Smith at CCRMA. (Most of them were mentioned
earlier.)
A very good source for the explanation of simulation by mass-spring models
and also modal synthesis is the book "Representations of musical signals"
by Giovani di Poli (ed.), published by MIT press. There are two (or three)
chapters by Claude Cadoz et al. (mass-spring and comparison of waveguides,
mass-spring and modal synth.) and J.-M. Adrien (modal synthesis). I can
highly recommend the article of Claude Cadoz et al., you can find any
formula you need for programming inside.


(Joerg.Spix@informatik.uni-oldenburg.de)