Re: structured looping interface (was Folk/Celtic Looping)

[Kim Flint <kflint@chromatic.com>]

At 10:42 AM 6/9/97 -0500, Dennis W. Leas wrote:
>Kim Flint wrote:
>> 
>> What do you think would be a more useful interface for real-time looping
>> with more structured music? Any ideas of what would work better for 
>you? (or
>> anyone else?)
>
>Yes, that's the $1,000 question.  I wish I knew the answer.  I've some
partly developed 
>(half-baked?) ideas.  Ever since your posting on "Looping and Composing"
I've been thinking 
>about this.  I want to put some of these ideas together in a coherent
(hopefully) form.  In the 
>meantime, here some less coherent writing:
>
>1) I use two EDPs to given me more control over the structure of the
compostion.  The problem is 
>shifting control parameters, such as 8ths/Beat, Sync, etc from song to
song.  I agree with Mr. 
>Dresher's comments - 
>
>"In my work, the ideal way to make structured
>composing/performance on these units more practical would be to simply 
>have
>these various parameter set-ups stored as programs which could be recalled
>with a MIDI program change, just like storing synth or DSP programs."
>
>I think I can solve this problem with something like the Peavey PC 1600.

Yes, we agree with that too. We are planning to add presets in some future
software upgrade. (we're a long way from releasing that, so don't go askin'
yet. :-)  The standard echoplex hardware only has enough non-volatile 
memory
for about 4 presets, which isn't so great. We'll probably make the software
support a sort of hardware upgrade, where you can change a part on the 
board
and get more preset memory. That will allow you to make rapid parameter
changes with program change messages.

As it is you can program the echoplex on the fly with a sequencer, and have
sequences that make particular parameter changes, but that's probably a bit
awkward.


>2) In my structured compositions, for a given piece I always tap the same
footswitches in exactly 
>the same order.  I don't want to use a sequencer to do this; I may choose,
in the course of a 
>performance, to vary how many times I play with a particular loop, among
other reasons.  
>Therefore, I think of using a single footswitch, preprogrammed to "walk"
through a cycle of 
>functions with each press.  This device would really be a front-end to
multiple EDPs.  For 
>example, in the schema for "Campbell's Farewell to Redgap" (a portion 
>below)
>

>
>The "preprogrammed single switch controller" would function something 
>like:
>
>time ->
>single switch:         1st press       2nd press       3rd press       
>4th press
>
>EDP commands sent:     REC             REC             OVERDUB         
>OVERDUB, MULTIPLE
>
>etc.


There are midi foot controllers that can do this. My Digitech PMC-10 has a
sequence mode, where a given bank is actually a sequence of patches. You 
can
increment or decrement through the patch sequence by tapping a pedal. It
even lets you increment or decrement with out sending the patch, so you can
jump to another part of the sequence. The patches can each be fairly
complicated midi strings, so this is a potentially very powerful thing. (I
haven't used this feature yet, but it's another reason why the good old
PMC-10 may be the best $100 I ever spent.) You can find them used 
reasonably
easy, I'd recommend it. 

It is a really interesting idea for a more structure-friendly loop device, 
BTW. 


>3)  A programmable "assistant" (Here's a real "blue sky" idea, I hesitate
sharing it.) - This 
>unit would have selectable or programmable "feature recognizers".  For
example, I could set it up 
>so that, when I touch a particular button, it means "I have a theme in
memory 1, with variations 
>in memories 2, 3 and 4 (or loopers 1, 2, 3, and 4), now generate
theme-and-variations".  I'm 
>thinking of some type of programmable AI rule-set that, when particular
features are recognized, 
>cause particular actions.  This could be something as simple as Reich's
phasing approach or 
>considerably more complicated.

I'm really interested in that sort of thing, but it's pretty complicated
technically! Especially in the general case, where somebody could do any
possible thing, and the device has to react in the way they want it to. a
tough one.

Have you ever used Opcode's Max? There are neural net objects available 
that
you could train for purposes like this. It takes a bit of effort to get it
working right, but the results can be very impressive. You could probably 
do
some pretty cool max-based theme and variation patches without the neural
nets too. You might try exploring that as a front end to the echoplexes.

They do a lot of research in this area at CNMAT. (center for new music and
audio technology at UC Berkeley, http://www.cnmat.berkeley.edu) Prof. David
Wessel is a big proponent of AI approaches to interactive musical
instruments. you might want to check and see what they are up to in this 
area.

kim