RE: question

[Greg Meredith <gregmer@MICROSOFT.com>]

Michael,

i have often faced the problem of lack of inspiration. i appreciate almost
all, and employ many of the techniques already suggested by many on the 
list
to overcoming this situation. However, one approach stands out by its
absence on the growing list of suggestions. This is the one that, for me,
absolutely hands down, conquers this problem. 

i explore the "design space" of sonic organization in an entirely
algorithmic and *mechanical* way. i decompose the space of sounds into a
finite set of dimensions and rules for combining sounds along these
dimensions. And then, i simply start exploring, in as methodical a manner 
as
i can, the applications of the rules to the generation of sounds.

What this does is to guarantee that i get into regions of the production of
sound that my aesthetic or gut-instinct almost certainly would *not* have
led me. This almost invariably leads me to a surprise: a combination of
sounds i would never have thought of putting together in quite that way.
When i get surprised, i get intrigued. When i get intrigued, i get 
inspired.
This is my experience.

Let me give an example. But, because the example is long, let me stress 
this
is an example of a general technique i'm trying to get across. Suppose that
we simply want to explore the production of sound in a single 'key' in the
sense meant in western classical music. One observation is that all tones
are naturally divided into 7 tone classes. Each class contains all the
different octaves of a given note. Thus, in a key, say D major, which has 
an
A natural, one of the tone classes contains A440, and A220 and A110 and
A880, etc. Let's write [A] for the tone class that contains all the A's and
[D] for the tone class that contains all the D's, etc.

Now, suppose that we assign to each tone class a number. For definiteness,
we'll stick to the D major key and, for simplicity, lets have them ascend 
in
a manner similar to the way the tones in an octave ascend. So, that gives 
us
a little table like this:

[D]    <---->  0
[E]    <---->  1
[F#]  <---->  2
[G]    <---->  3
[A]    <---->  4
[B]    <---->  5
[C#]  <---->  6

Given a tone class [X], let's write N([X]) for the number we assigned to 
it.
E.g., according to our table N([D])=0. Similarly, given a number, n, 
ranging
from 0 through 6, let's write T(n) for the tone class assigned to it. So,
T(0) = [D]. Notice that N(T(n)) = n and that T(N([X])) = [X]. For instance,
N(T(0)) = N([D]) = 0 and similarly, T(N([D]))= T(0)=[D].

What we're going to do is to introduce a new kind of transformation on
collections of tones. It's sort of like transposition, in the sense that it
preserves some symmetry, but it's not transposition because a) we never
leave the key and b) we preserve a different symmetry than transposition
preserves. 

To introduce this transformation, we're going to pair up a number from 0
through 6 with another such. Whenever any two numbers between 0 and 6,
inclusive, add up to 7, they're considered a pair. So, the pairs are

1,6
2,5
3,4
4,3
5,2
6,1

Now, i'm going to throw in the pair 0,0 to complete the story. You can view
it that 0 was the only one not paired, or you can view it that we're
operating in the group Zmod7, whichever works for you. Given a number, n,
from 0 through 6, let us represent it's partner by p(n). For example, p(6) 
=
1 and p(1) = 6. Notice that p(p(n)) = n.

Now, we can write down our transformation. Given a tone class [X],
Rotate([X]) is defined by Rotate([X]) = T(p(N([X]))). Here are the 
rotations
of all of the tone classes for the key of D according to this assignment.

Rotate([D])= T(p(N([D]))) = T(p(0))= T(0) = [D]
Rotate([E])= T(p(N([E]))) = T(p(1))= T(6) = [C#]
Rotate([F#])= T(p(N([F#]))) = T(p(2))= T(5) = [B]
Rotate([G])= T(p(N([G]))) = T(p(3))= T(4) = [A]
Rotate([A])= T(p(N([A]))) = T(p(4))= T(3) = [G]
Rotate([B])= T(p(N([B]))) = T(p(5))= T(2) = [F#]
Rotate([C#])= T(p(N([C#]))) = T(p(6))= T(1) = [E]

Or without the intermediate calculations

Rotate([D])= [D]
Rotate([E])= [C#]
Rotate([F#])= [B]
Rotate([G])= [A]
Rotate([A])= [G]
Rotate([B])= [F#]
Rotate([C#])= [E]

Now, suppose you give me a melody in the key of D major. If i faithfully
apply my transformation to the melody, that is, i rotate all the notes in
the melody, then i get an entirely new melody still in the key of D major.
All the rhythmic values have remained unchanged, but the note-value contour
of the melody has completely changed. 

You may have noticed that i only gave you tone classes, not tones. So, to
apply rotation to a melody, you still have to pick one of the tones from 
the
tone class to get a real tone. But this is one of the interesting 
parameters
to vary in this process.

Another parameter to vary is the assignment of tone classes to numbers. You
could, for example, try

N([B])=0
N([C#])=1
N([D])=2
N([E])=3
N([F#])=4
N([G])=5
N([A])=6

And any of the other many assignments. 

So, popping up a level we now have a rule for generating new melodies from
old ones. The next step is apply this rule as methodically as possible,
trying to come up with as many ways as you can to exhaust its application. 
i
guarantee you, you will be surprised by the results. 

The MIDIscenti on this list might write a program which applies such a
transformation to a loop. A really cool approach is to put the 0 on the
starting note of a melody, apply transformation and loop that once, then 
put
the 0 on the second note of a melody, apply transformation and loop once,
etc. 

Philosophically, i believe that this meta-technique of exhaustively
exploring the design space generated from a few simple rules is exactly 
what
happens in nature. Natural selection speeds up the truly exhaustive search.
Similarly, in this technique we're, effectively, employing a genetic
algorithm where 

*       the individuals are 'musical utterances';
*       reproduction is the application of a rule;
*       the fitness criterion is 'do i like the way it sounds.' 

It also pleases my sensibilities no end to derive inspiration from an
entirely mechanical process. It fits with my view of the mystery of life: 
so
much awe-inspiring variation from just enough mechanism to make it happen.

Finally, it takes away most of the ego from the process. It's very little
about me or my generative and potent creativity. It's more about my tiny,
limited aesthetic (which was more or less developed before i was even
cognizant of 'i') reacting to and interacting with the vast array of
information that's already there in front of my nose. And, when my
impoverished imagination fails to help me see what's right in front of me,
turning the handle on the machine will.

Sorry the post was so long. i didn't have time to think about how to make 
it
shorter.

--greg

        -----Original Message-----
        From:   Nemoguitt@aol.com [SMTP:Nemoguitt@aol.com]
        Sent:   Monday, January 25, 1999 11:01 AM
        To:     Loopers-Delight@annihilist.com
        Subject:        question

        what do you all do when your muse goes on vacation?.......i have 
hit
a wall
        and it seems all i am now doing is attempting to make "silk purses
out of sows
        ears".....throughout december i was a nova of creativity, burnt
miles of
        tape.....now for the last week or so i cant even buy a new
idea.......any
        methods out there to re-kindle the spark or is it just a matter of
        waiting?...........michael