Sibelius

Hello tuning list people,

I have been reading david Cope's books, about musical machine composition
experiments, with great interest.

Yes, Bill Alves was correct, he doesn't composed COMPLETELY from "thin air".

However, there two are basic methods of composition, which I have identified:

1.PARASITIC COMPOSITION: This uses music composed already by a human, and
either arranges it, rearranges it, or manipulates it, to give the false
impression that it is a new composition.

We can have a LOT of fun here, using life (cellular automata), ants, matrix
propertys ect.....

We can also do this STATISTICALLY, by analyseing the pitch, duration,
volume, and temporal placement of the notes in the score, on all time
scales, and using a population sample formula:

q=E((m-p))^2/(n-1),

where:

q=Standard deviation, E=Sum of ala "Sigma notation", m=Measurement, p=Mean
(Sum of population elements/number of population elements), n=Sample size.

Many samples can be taken, of all sizes and resolutions, on time, pitch,
volume, and note duration scales, and these can be multiplied by a
coefficient for bias, and used to compose from a random number generater,
which could change via time, or, we could pool all the sample standard
deviations togther and take the secondary standard deviation, again,
multiplying n-1 by another coefficient, of course to get rid of the effect
of the other coefficients.

We can have a "fractal analysis-creation" algorithm.

I really can't say what the results of this would be, but they should make
for interesting experimentation.

Interestingly, no matter what the characteristic of the nature of the
population, we always get a rough bell curve normal distribution shape, when
appling the standard deviation formula.

I intend to change the formula to see the results.

2.LOGICAL ALGORITHMIC COMPOSITION:Use a set of rules, directly from
mathematics, rules which are inherent by the very nature of number, and are
JUST THERE.......

These can be:

-Chaotic,
-Brownian,
-Fractal brownian,
-Formula roundoff feedback,
-Ant in a matrix,
-Prime spiral,
-Snowball/hailstone,
-Formula,
-Stochastic,
-Genetic,

ect.....

This brings to mind an interesting point about number theory.

"Are the numbers RANDOMLY distributed-the propertys of which we perceive to
be inherently "logical", and having "Meaning, and depth", just man made"?

Ramsey theory states that we will always find some sort of order in any
random set of data, so, this being the case, are events around us just
"random", and what we percieve as order an illusion?

---It would very definitely take a lot of time to input all not values from
a staff and clef score of music, so, I was looking at buying a scanner and C++.

There is comercial software for reading scanned scored music, and playing
it, available on the market, namely "Sibelius", but it is a tad expensive.

To get around this, I have gone to the library, gotten out a copy of the
Rudiments of music, with most of the musical symbols you'll ever find on
scored music, and I intend to write a primitive programme for emulateing
Sibelius, from the digital data taken from a scanner.

Scored music is inherently deficient in information---a far, FAR better way
of diplaying a composition, is piano roll.

However, the lack of data available to us in scored music, could actually be
seen as an advantage, really, the fact that we don't see EVERYTHING means we
can improvise, and lend character to our performance.

I also will be experimenting with a humanizer/humanizeing algorithm, to
roughen up the output a tad.

(One of the reasons I dislike mainstream techno-too robotic and cold).

Oh, yes, and the programme will have a wide selection of microtonal capabilitys!

Any thoughts?

Sincerely,

Sarn "Polyverse Philosopher" Ursell.