Something like markov-chains but w. index...
So, I've been having a great deal of fun with anal and prob, but... For my application I need something, which also takes an index. That is, something which is fed a series of numbers, toghether with an index (16th notes etc) and builds up probabilities tied to that index, so when queried for that index point, it puts out numbers tied to the index.
Why?
Well, markov chains are fun, but it becomes unmusical as the series quickly becomes dislodged from the rythmical patterns that were originally in the stream of numbers.
Any ideas?
Are you talking about an array of markov chains? Where each index
would point to a state table?
Yeah, kind of, but the choice has to be dependent on what choice was made previously, in the step before that...
For example:
im7 ivm7 iim7b5 V7b9 im
I7b9 ivm7 bVII9 bVII7 bIIIMaj7
bVIMaj7 iim7b5 II7 V7#5#9
im6 ivm6 V7#5#9 im6
im6 ivm6 V7#5#9 im6
im7 vim7b5 iim7b5 V7#5#9 I7 ivm7 bVII9 bIII9
bVI9 iim7b5 V7#5#9 im7 vim7b5 iim7b5 V7
I7 ivm7 bVII9 bIII9 bVI9 V7#5#9 im7
I7b9 ivm7 I7b9 ivm7
I7b9 ivm7 I ivm6 I7b9
ivm6 iim7b5 vm7b5 I7b9 IV7 bviim7 bIII9 bVI9
bII9 bII9 bVI9 bVI9
bII9 bII9 bVI9 iim7b5 V7#5#9
These are the chords in Django, as you can see there are reoccuring patterns split basically into four "columns", with occasional subdivisions, I want to have a first order markov chain matching, which takes the current column-number (1-4) into account, not just going from chord to chord at random.
So, upon further studies, I've found out I need a looping index of 16, with a 2nd order.