Digitals--

[John Musselwhite]

At 10:20 AM 2/15/98 -0500, Stephen wrote:

>To clarify here, when I talk of the next generation these will not be
>"digital". With "physical modelling" the input/output response itself is
>modelled. Particular waveforms are not linked to particular input
>combinations (interpreted from key motion). No matter how many of these

This is what Sound Blaster is attempting to do with the Advanced Wave
Effects (AWE) system they're using in the SB64 AWE and other cards. The
technology uses what they call "sound fonts" that sit in RAM (4 megs of
it!). The piano sound is still a little weak, but it's miles ahead of what
it was on earlier sound cards.

I have a 49 key "Quick Shot" MIDI controller connected to mine and it works
great for what I'm doing, which isn't anything fancy. It's certainly an
inexpensive way to get into MIDI. 

>my reference to "next generation" as the keyboards would bear no
>connection with sampling types, such as even the best are currently. It
>will be quite possible to mimic the real piano closely (at least for mass
>consomption) by this method. First we need to learn a lot more about how 
>the real thing responds.

Well, we DO know a lot about how the "real thing" responds... that's part
of the problem! The biggest problem with physically modelling a piano is
the complexity of the interaction of the strings and soundboard as well as
the hammers and strings. While certain things can be "fudged" in the model
there will always be things they have to leave out simply because there is
too much going on for a model to handle. Even with 1,000 mhz computers with
a gigabyte of RAM the model will have a lot of compromises, and it will
always be just a "model" of the real thing.

I remember reading a Star Trek novel one time where a concert was given in
10-Forward on a Steinway grand piano. They didn't specify whether it was
created in the "replicator" or in a real factory, but the point was that
even in the 24th century concerts were (?) given on a "real" piano, not an
electronic one.

		John


John Musselwhite, RPT  - Calgary, Alberta Canada 
musselj@cadvision.com - http://www.cadvision.com/musselj/