On 8/13/06, Marius Schebella marius.schebella@chello.at wrote:
Chuckk Hubbard schrieb:
Two things come to me right away. One of the things Bach did was to start using certain notes that implied a key change before there was any indication of momentum towards a key change. It's easy to work a Bb into a theme in C without giving too much away. Then when it happens, the change sounds like it fulfilled some subtle destiny. I have *no idea* how you could do this with Pd.
if you use Bb in C, it is another chord, C7 (C - C7 - F)
I have no idea how to make a patch that would insert pivot tones somewhat randomly. I like the neural net idea, where it just learns certain things according to how well a user likes them, but I have no more clue how to make that than the other. Could it be made to notice similarities it wasn't specifically told to notice?
You can get away with a lot of little borrowed progressions with jazz. Any chord whatsoever can be preceded by a dominant 7th, or a tritone sub for a dominant 7th, and that in turn can be preceded by a minor 7th, or a tritone sub for that, and so on. You could actually have lots of momentary forays that would end up returning to the original key.
If you have a 4-bar period, and just just use a II-V-I progression like |Dm7 |G7 |Cmaj7|(Cmaj7)| (|Xy| = one bar) these harmonies can be substituted in many ways, like |D7|Db7|C6| or |D7 - Ab7|G7 - Db7|C| with two changes per bar. OTOH the other hand you could use substitutions for major harmonies by their parallel minor harmonies |Dm|Em|Am| and of course also the other way round |F7|B7|C|. some harmonies are weaker, some are stronger, so define, at which time you want to have stronger changes in harmony and when there should be little changes. then have other progressions, styles, tastes... m.
Of course, if all movement is by fifths and seconds, it will defeat the original purpose, which was to make music that was less predictable. I think there'd have to be some movement by thirds, and just plain using chords that aren't in the same key. I'd actually press for extended just tuning, if I were doing it (I'm not). Damian, maybe it would be possible to simply have a melodic part that occasionally receives information about harmonic changes before the harmonic part does. I like your idea, but I think it would take quite some time to realize it. I haven't checked out the Alexandre Quessy project mentioned above. Maybe a project like this would be more for doing it than for having it, though, as so many computer music projects.
-Chuckk