additive

[The Dark force of dance]

Y-ellow Y'all.
	I'm only breezing through and I've not caught up with the rest of this
thread. But!

The process is called "RE-SYNTHESIS" There are loads of tools available to
do it. From simple software programs on a PC through to synthesis systems
such as the creamware scope.

Indeed. it uses a fourier transform to break down the original sample into
it's component sinewaves (harmonics) and assignes them to their envelopes
over time. The more transforms (sinewaves) and the more detailed the
envelopes, the more actuate it will reproduce the sound.

Once you have this data and the means to reproduce it, there is very little
you can't do with the sound. At the very least you get a sampler that
allows you to play the same sound at any pitch and maintain the same
transient. At it's best you can gene splice seemingly unrelated sounds
together. A chromatic cymbal springs to mind. A drum cross with a piano
perhaps. Or simply encode an entire piano using just 9 samples. Instead of
the thousands it takes to do it using the usual brute force method.

But the thing to note here is that all these systems that do this remotely
well are "DIGITAL". It is very difficult and cumbersome to do this stuff
using analogue. The Kawai K5000 for example, has 6 layers. Each layer
consists of 64 sinewaves. Each with 64 envelopes. Plus filter. And there
are 32 voices of it at once. Try doing that with a modular. I mean a
modular that you can fit inside your house instead of a 4 story office block.

The K5K also comes with knobs so you can adjust stuff in real time. But
before I sound like an advert for Kawai, it should be pointed out that they
don't do re-synthesis as such. The envelopes are only ADSR types. These are
nowhere near good enough to do re-synthesis over all. For that your
envelopes need to be a uni-polar sample of the harmonic's amplitude. It
doesn't necessarily have to be particularly high resolution in either the
time or amplitude domain but it does have to be significantly better than
that of which a simple ADSR can provide.

FM is another matter entirely. FM synthesis is only classed as "additive"
because it's certainly not "Subtractive" This could be viewed as a lateral
form of synthesis. Or perhaps recombinant harmonic synthesis. At first
glance there seems to be no direct relationship between any two operators.
The harmonic content of any given operator is pushed but the phase of any
operator preceding it. In order to make it "re-synthesize" some very tricky
software is required. Yamaha's second foray into FM synthesis, the GS1 and
GS2, did just that. Only what you had to do was send a tape to Yamaha
containing the sound you wanted encoded. They would take that sound and run
it through a fairly time consuming process which would reduce it to control
data for their (3 operator) FM synthesis engine. Finally you'd get a little
bus ticket thing back which you could insert into your instrument and it
would do a half decent job of replaying the sound. (Of course Yamaha's
first foray into FM was on the some what funky GX1. 2 operator analogue FM)

Whilst it should be theoretically possible to back-calculate a sample into
an FM control set for a DX7, to my knowledge, no-one has actually done it.
But it should also be noted that a DX7 has better envelopes than a simple
ADSR type as well. Which is one of the reasons it's better at simulating
the noiances of real-world sounds. Possessing 8 parameters instead of 4, it
allows for harmonics to swell. In subtractive you need to dedicate an
entire chain to do it.

In short. If you wanna do this stuff, it's possible to do it using an
analogue subtractive system but why would you want to?

Be absolutely Icebox.