adsr and lfo abolish

[Martin Czech]

Ok, it seems I did start a new thread.

I'm sorry that I made my point not so clear, though most of you have
understood where it was going.

Ok, true: first it deals with the kind of music one does. 
Don't worry, I won't get into a musical discussion here, just stick to the technical aspects.

For classical pieces in the manner of "Switched on Bach" the
ADSR-Key-Trigger concept did work apearently very well.

But it is possible to think in another way, a music that can not be
described by means of trigger events, ie. specific points in time, where
modulation envelopes start. Music is possible where modulation parameters
change arbitraily over time, with no restrictions. Most polyphonic keyboards
can not do this, because they start all envelopes of a voice at once if
the trigger comes. A modular system could to this with a lot of ADSR and/or
LFO circuits, that act independently. But this would be a lot of pain
and one would have to record "event" by "event", because the settings of the
ADRS/Trigger-Machine would have to be readjusted for the next second.
I know that some ADSR designs can be retriggered, this helps a little bit 
for the synthesis eg. of flute like sounds, but the general problem
of arbitrary envelopes can not be solved this way.

It was said that Moog's decision to use a keyboard was the key to success,
and I believe this decision was very influential up to the presence.
Nearly all machines have keyboards (or midi) and allow modulation ONLY by
envelopes or lfos. The midi standard was consequenly made to primarily
transmit key-on, key-off information, well and some controller action,
we know that there are problems with 7-Bit limitations and bandwidth.

The string section example was maybe a bit misleading, I only wanted
to stress that even in conventional music "envelopes" are very variable
whereas most synthesis machines use rather fixed shapes. The ear is very
good in analysing the resulting spctral changes und thus will very soon
find out, that every event is similar to the next, thus this kind of
fixed envelope treatment can be boring.

A straight forward solution would be to use a pc and a set of DACs with about
audio rate in order to generate the necesarry modulation voltages. But
this would be ridicolous, how would one program these curves? It would
be possible to store these however, since a few GB don't cost much
today. But this would mean to start from scratch.

Another way for implementing arbitrary control voltages is to use
midi. This would have the advantage not to have to start from scratch.

Now, it was pointed out that there are bandwidth problems, and also
bit-resolution problems. (I was almost shure of this, but I did ask in
order to be very shure. Midi II is on the way for years now, I don't
know if it will ever appear on the market. And I fear that the above
mentioned modulation problems will not be attacked, since 90% of the
musicians would not even know what I'm talking about here.)

The problems could perhaps be removed if the arbitrary envelopes would NOT
be transmitted in a straight forward sampling approach but in a continous
simple reprogramable envelope generator/filter manner. Ie. clever
interpolation of some few data points. This way only some key parameters
have to be transmitted, eg. time, asymptote level, shape and starting
time.  In order to avoid latency time it would be possible to program
parameters first and then set them into action by a key-on message.
This would lead to continous control voltages, but would require a midi
interface that interprets midi information in a quite unusual way.

I don't think, that such a midi interface is avaiable on the market.

By the way: are there any controllers (not sysex) with 14 bit resolution
other then extended pitch bend? I know of none.


On the software side it would require a tool (interpolation coding
software) that translates drawings or other specifications of the
arbitrary envelopes into interpolation data, in order to feed the midi
interface.


How would a good analog interpolation circuit look like ?  With regards
to the simplicity of the interpolation coding software.


A lot of pain, but I think very interesting results can be obtained with 
changing just a few parameters arbitraily over time.

So far I have only experiences with the Micro-Wave, these are very
limted of course because only four 7 Bit controllers where involved (X,
Y, Z, W) and they could only change in a very slow manner (sliders are
not that fast).  But even with such strong restrictions I could obtain
very interesting results, it was really a new sound , like an entire
new synthesis machine.

m.c.