le 2001/08/23 19:10, Richard Goodman à bhomme@... a écrit :

> I don't immediately see how one would
> apply "Greeking" to music composition.

no it wasn't greeking but the same algorithm that i used to 'greek'.

i will explain slowly.

a melody can be defined as a sequences of notes, each having a pitch and a
duration, pauses being a non-note of specified duration. i will not here,
for obvious needs of simplification deal with the notions of 'instruments'
nor of the expression of the musician [a wonderful article in 'wired' this
month about ai research into musical expression]. nor will i deal with a
real performance of that melody which would also need factors such as keys,
volume, arrangements, different instruments, expression, mixing...

for the purpose of this explanation any melody can be transcribed then as
note+ duration. it can thus, with a minimal change to the algorithm, be used
as a 'source', and parsed to produce other melodic phrases. the greater your
source, the better melodies you produce.

another approach would be to not treat note+duration pairs, but seek these
out separately from the same sample source, and then pair then at output
time.

i don't see where the problem is in changing the source from fondamental
language [or word] elements, to melodic [or note+duration] elements. the
fundamental algorithm is the same.

:-j