GG: Sampling CD318

["John P. Hill" <jphill@frank.mtsu.edu>]

On Fri, 31 Jul 1998, Huw Davies wrote:

> Whilst walking from one disaster to another at work today (I'm not quite
> sure why the two areas that I supervise are separated by a Km or so...) I
> wondered how hard it would be to "clone" CD318 using modern sampling
> techniques. I know reproducing the feel would be difficult, but it would be
> really neat to have something to download into an electronic device (piano,
> computer, etc) and try to make sounds similar to GG (humming would be up to
> the performer however :-)

Yeah, that's a pretty tall order.  Sampling technology is now quite
advanced, but I think this would entail a good bit of work and at the end
of all of it, you might just stand back and go "why"?

You'd certainly have to do *multisampling* of each key, so that a range of
dynamic effects could be obtained.  Part of the difficulty in sampling
real acoustic instruments involves how you map *performance hooks*
(timbral variations with velocity, etc.) with the sampling hardware and
software you have at hand.  One of my favorite keyboards to play in the
studio is the Hohner Clavinet (remember Stevie Wonder's SUPERSTITION and
HIGHER GROUND?).  I've not yet heard a really convincing sample of it,
partly because it's pretty hard to map subtle pitch shifting and
distortion content to key velocity and that's something that happens quite
organically when you hit the clav's keys hard.  So you have to know *what*
to map onto the various performance modulators and then *how* to model
those effects.  Not trivial.

Beyond this performance hook problem, you have the issue of the tactile
aspects of the keyboard itself.  It's just very different playing on a
plastic (even weighted) keyboard vs. the actual keyboard itself.  With a
real piano you have that wonderful sense that your fingers are actually
connected to a hammer and a string (ie:  the actual sound-producing
mechanism).  Playing a set of samples is always a step down, IMHO.

It would be best, of course, to actually "sample CD318's DNA" directly
(sorry if I sound like Mephisto on SOUTHPARK), althought this
multisampling process would be quite time-consuming and tedious if you
were to do it right.  On the other hand, it might be possible to do some
of this sampling from existing recordings.  Don't tell Mr. Sony.  It's
doubtful that anyone would come after you for lifting individual notes to
your sampler, but strictly speaking, this probably would be an 
infringement of copyright.

jh