Add algorithm / analyzer to improve voice allocation in generated material

[ciacob]

Add algorithm / analyzer to improve voice allocation in generated material

Currently (1.5.6) voices are picked purely randomly, unit by unit (which, most of the time translates to "note-by-note") when generating music. This results most of the time in sub-optimal voice allocation, to the point where everything is way to scattered (like in Webern's music), and no voice / instrument plays a clear line.

I intend to change this by adding an analyzer that provides context-based scores to proposed notes, the way that a voice that merely begun playing is encouraged to continue to the disadvantage of one that is currently silent.

There are more such heuristic rules, which are described in the attached diagram (Romanian only for now).

[ciacob]

voice-allocation-draft.drawio.pdf

[ciacob]

voice-allocation-draft.zip

[ciacob]

This is a very old draft, most like suggested by ChatGPT, and most likely unusable, but I archive it as such, anyway.
algorithm stub for voice distribution.txt