Thanks for the feedback (and compliments)
Yeah, I think I understand what you mean.
To make myself clear, I'll go into a bit more detail about how the structure is generated:
I tell each 'voice' how many bars I want it to fill, then I tell it when to start and how many times to repeat. It then generates a 'structure'. For example, I may tell my melody to generate a structure that's 4 meters long, and repeats three times. In which case it might generate:
[A1, A2, A1, A2] x 3
The letter representing the rhythm, and the number representing the melody to go with the rhythm. So basically, this structure would repeat the same rhythm four times, with a single variation on the melody.
The repetitive-ness is why I think you find it predictable. Unfortunately it only "goes somewhere", as you put it, because I programmed it to 'build up'. This was mostly because I didn't figure people would give it more than one listen, and I didn't want to scare people off by starting off with as much complexity as it ends up with.
Right now it's not at a stage where it can transition from one 'part' of a song to another (eg. chorus to verse), nor can it generate particularly meaningful structures, with the right amount of progression vs. catchiness, on it's own (hence me setting it to repeat). If I tell it to generate, for example, a 12-bar structure, it might generate something similar to:
A1, A2, B1, A2, B1, A2, B1, A2, B1, A2, B1, A2
Getting 'stuck in a loop', essentially (due to the way I programmed it). It is difficult to generate a balanced structure (ie. it's catchy but not boring) - especially since there are four voices to balance - but that'll be my next task. Any suggestions would be helpful (as there's not a lot of music theory referencing how to structure compositions).
I can change how likely the note lengths are to vary from the 'average' I've set - I'll mess about with it when I get back from my short holiday on Sunday.
In fact, I was thinking of side-lining this project, however if you guys think it has some potential, I might carry on with it in parallel to my other (as yet undecided) project... I may give it has a GUI, meaning I can distribute it - that way, you guys can play around with some of the values used to generate the music.
As for the MIDI files...
Here's "Output Ten":
http://dl.dropbox.com/u/10171005/output_ten.midAnd here's a newly generated one, just for you:
http://dl.dropbox.com/u/10171005/output_eleven%20%28new%29.midSorry all the voices are in one file (you can work out which voice is which by which octave it's in, and at what point the voice comes in) and that it's so long/repetitive, but I'd be really happy to see you upload a version with different instruments
EDIT: Also, "Output Ten" is written in 3/4, so you might have to account for that in your soft-synth, so it doesn't display notes all weirdly