R: [PD] pd-based procedural chord progression database..

[Davide Morelli]

you may want to check a couple of externals I made together with David
Casal:

[chords_memory] 
and
[harmonizer]

[chords_memory] takes as input chords names (as given from the [chord]
external from maxlib) and is able to learn your style while you play. when
you have trained it a bit you can ask for chords sequences in your style.
it stores the probabilities of transitions between chords in the current
tonality in a weightened graph.

for example if you train it with this sequence:
in the C major tonality,
C maj, D min, G 7th, C maj, F maj, C maj, F maj, G maj, C maj
you can then set the current chord to C maj and ask for a next chord with a
high probability and it will tell you F maj, if you ask for a lower
probability it will tell you D min.
you can change tonality. you can also store the graph in a file for later
usage.

[harmonizer] creates voicing for you:
given the current chord and the current voices in MIDI values it attempts to
create a nice voicing to a target chord. you can set some parameter like how
much you like parallelisms, etc.. it can handle as many voices as needed but
you must tell it the number of voices at creation time (as the first
argument). it uses genetic algorithm but i noticed that the optimal solution
is found at the first generation.

they are in CVS in
/externals/frankenstein

i'm sorry i didn't wrote any help patch yet (no time so far), if you are
interested ask me for more details and i'll be happy to help

ciao,
davide.


> -----Messaggio originale-----
> Da: pd-list-bounces at iem.at [mailto:pd-list-bounces at iem.at] 
> Per conto di Damian Stewart
> Inviato: venerdì 11 agosto 2006 7.15
> A: PD-List
> Oggetto: [PD] pd-based procedural chord progression database..
> 
> hey everyone
> 
> i'm setting out to build an automatic chord progression 
> system - you tell it to start in C Major, centre around E 
> minor, with a given deviance factor, and then bang it 
> repeatedly and it spits out a new chord/voicing each time you 
> bang it, that makes harmonic sense and sounds good as a progression.
> 
> (why? it struck me while i was listening to my friend's set 
> the other night that too much dance- or dance-derived music 
> is harmonically static. simple
> solution: make a quasi-intelligent system that knows what 
> progressions sound good, and get it to do all the work. 
> imagining coupling it up to fiddle~ and a vocalist..)
> 
> based on jazz harmony theory, which i have had the tiniest of 
> nosies into, this will involve encoding knowledge about all 
> twelve scales, major and minor, as well as the modes and 
> relationships between them. before i start, is there anything 
> i should look at that might help, or solve the problem for 
> me, or make the whole idea irrelevant?
> 
> cheers
> d
> 
> --
> f r e y
> live music with machines
> http://www.frey.co.nz
> 
> _______________________________________________
> PD-list at iem.at mailing list
> UNSUBSCRIBE and account-management -> 
> http://lists.puredata.info/listinfo/pd-list
>