First, for those folks interested in learning more about granular synthesis,
there is a good background article in the June 97 Keyboard mag, if you can
find the back issue.
Second, I'm not an expert and I'm going from <hazy> memory...
For granular, you want to create little "packets" of sound and string them
together, so you would need lots of VCAs and envelopes to shape packets, and
VCOs or other sound generators for the packet waveforms. These packets are
often just a few milliseconds long, and strung one after the other. You
would need delay on your envelopes, and probably the whole thing would be
easier with purpose-built envelope generators that were set up to allow very
precise control at the very short end of the time spectrum. Granular also
has the concept of "grain density" - I'm a little fuzzy on how this actually
works, but I think you'd simulate this by bringing in more signal chains in
parallel so that they sound simultaneously. Anyway, lots of VCO, VCA, very
fast EGs.
This is not the same as wavetable synthesis. A few of us had this discussion
offline yesterday, but I'll repeat it here. A wavetable oscillator stores
many banks of short looped samples. It can then be set to play these loops
independently for timbres that give you something different than the
traditional analog waveforms. You can also play backwards and forwards
ACROSS these different banks under voltage control, giving you a new way to
control shifting timbres for some nice evolving sounds. Think PPG, Prophet
VS, Korg WaveStation.
Personally, I don't think granular is that practical on an analog modular,
but it's certainly possible. On the other hand, a MOTM wavetable oscillator
would be an awesome addition to the sonic arsenal, especially when you can
run it through all those killer filters we're going to have soon. It is a
much simpler and less hardware intensive way to get lively animated sounds.
I hope to see a wavetable module at some point.
DX-7 style FM synthesis is more straightforward than granular. Again, no
filters are needed (although you could certainly add them to the mix), just
VCOs, EGs, and VCAs. Use sine waveforms on the oscillators to emulate the
DX-7. You need a VCO into a VCA, controlled by an EG for each operator (a
DX-7 had 6, the smaller TXs had 4). Then just patch your "operators" like
the patch algorithms that were printed on the DX-7's front panel. It's
basically one oscillator (the modulator) modulating another (the carrier)
through the VCA. Use the FM1 input, and linear mode will give you more
controllable results. Because you are controlling the gain of the modulation
through the VCA, the EG becomes the "modulation index envelope". The
algorithms are just different parallel and serial modulation connections.
So, to imitate a monophonic DX-7, you need 6 VCO, 6 VCA, 6 EG.
Then there's pure additive synthesis. You just build your sound 1 harmonic
at a time with a whole mess of sine wave oscillators, each through a VCA and
with its own envelope. It would take DOZENs of oscillators to make a sound
that would begin to have harmonically interesting content. It's the same
problem as granular - you have to microscopically control too many things to
be able to create sound manually. This kind of sound creation almost has to
be under computer control.
Dave Bradley
Principal Software Engineer
Engineering Animation, Inc.
daveb@... > >
> >BTW, these alternate techniques such as granular and FM synthesis CAN be
> >practiced on modulars, given enough hardware.
> >
> Hi Dave.
> can you please explain more how this techniqe can be done in modular,
> what moduls needed ?