Microtonal AI/SARA

Dear People of the alternative tuning list,

I am looking to experiment with David Copes SARA (an artificial intelligence
type music generation program written in "LISP"), in relation to Xentonality.
SARA has fooled people, composing music in the style of Mozart, Joplin,
Bach, Brahms.

In a recent experiment with music lovers, using a a)student composing in the
style of Mozart, b)real Mozart and c)David Cope's SARA, most people guessed
the imposter was a).

I am looking to buy a PowerMac, and to convert SARA to microtonality-and
therein lies an arduous task.

What would perhaps be easier, rahter than learn a whole lot of theory, is to
construce/create a BIG chart of chords and intervals (common and not so
common), for each temperament.

I must put this to you:

Would it be as simple as using 12 Equal generated music, and randomly adding
say, 1/4 and 1/3 notes, or would we have to define the rules from scratch
for each temperament?
(This may be a better option.......).

In Just Intonation, assumedly, we would just have to round each 12ET note
off to the nearest N-limit JI interval.

For example: Below are 4 common triads fot 24 Equal temperament (1/4 notes):

5.5
|
2.5 4.5 | 7.5 9.5 11.5
| | | | | |
|- -| | -|- -|- -|- 12.5
|| | | || | | | | | | | | | | |
| | | | | | | | | | | |
-|R | --|R | | -| | -| | -| | ---
| | | | | | | | |R || |R || |R || |
| -- | -- -- | --- | --- | --- | |
| | | | | | | | | | | | |R |
-----| B |--| B |--| |-|B |--| B |--| B |--| |
| | 2 | |4 | | | |7 | | 9 | |11 | | |
| --- --- | | --- --- --- ---
| | | |R | | | | |
| W | W | W -- W | W | W | W |
| | | 5 | | | | |
| 1 | 3 | | 6 | 8 | 10 | 12 | |
| | | | | | | | |
-------------------------------------------------|***OCTAVE***
|
|
C Major=1st, 5th, 8th notes.
C Minor=1st, 4th, 8th notes.
C Halfway=1st, 4.5th, 8th notes
C Superhalfway=1st, 5.5th, 8th notes.

Regretfully, I cannot show you the red quarter note keys.

Of course there are the chord inversions, 7th chords ect, which I will not
mention here.

However, using SARA to compose in 12Et, would it be correct to "randomise"
the 4th-5.5th notes by rnd(1)*3?

Or, would this just be a little like translating English directly into
Zulu/Japanese/Danish, and not altering the grammatical rules?

And why not just take music already composed and "overlap" the alternative
temperament to the nearest 12ET note?

Would you have to add more notes for the higher numbered temperaments, eg
53ET, number of notes=(53/12)*n?

I would definitely like to analyse samples via time, and find their
"usefulness" via the temperament, eg.a sample of a synthesizer sound, having
11 partials with (1/(((2^(1/12)))^(13 right through to 24))) times the
amplitude and frequency would be considered as "perfect" for that temperament.
Correct????

Thus, how would irregular samples of guitars, elephants, Mandelbrot noise,
burps, and "The Ziltch" be rated via time?

Any thoughts?

Will be making a CD in mid-1999.

Sincerely,

Sarn Ursell.

On Wed, 27 Jan 1999 13:40:47 +1300, Sarn Richard Ursell
<thcdelta@pop.ihug.co.nz> wrote:

>Would it be as simple as using 12 Equal generated music, and randomly adding
>say, 1/4 and 1/3 notes, or would we have to define the rules from scratch
>for each temperament?
>(This may be a better option.......).

Anything that can be written in standard notation can also be played in 19,
31, or 43 notes per octave if it doesn't rely on enharmonic equivalents
that are characteristic of 12-tet (D#=Eb, for instance). These tunings
correspond roughly to 1/3-comma, 1/4-comma, and 1/5-comma meantone.

Some 12-tet music can be translated to just intonation, but there are often
many possible JI equivalents of a note (e.g., D=9/8 or 10/9, Bb=9/5 or
16/9), and choosing the wrong one may sound very bad (the "wolf fifth"
between 9/8 and 5/3 is 40/27, which is 21.5 cents flat compared with 3/2).
Besides, I think it's more interesting to take advantage of intervals like
7/6 and 11/9 which aren't part of the normal 12-tone vocabulary.

Based on my experiments with 15-tet, 16-tet, and 20-tet, if you want to use
some of the more exotic scales like these you'll probably want to develop a
separate music theory for each different non-12 scale. There may be
similarities between some scales. The 16 and 23 note scales both contain
"pseudo-diatonic" scales of 5 small steps and 2 large steps, for instance.
Both also have very flat fifths, close to the 40/27 "wolf fifth". So you
could probably use similar rules for both scales, although there would be
subtle differences. 23-equal doesn't have a very good approximation to 7/4,
but 16-equal does.