This is a neural network library that I coded in Lisp as a toy-project around 2008. I post it here just for historical purposes. I do not plan to mantain the library in any way.
This is a little thing that I have done some time ago and never really got time to polish and publish the way I wanted, but it is still nice to put it out there.
This scripts generates X decision processes with some parameters and transform them in music. This just answer to the question that, for sure, every one of you is asking himself: how does a diffusion process sound like? And the answer is: horribly! As expected 😉
To code it, I used the nice library of Ken Schutte to read and write MIDI in MATLAB.
In the code, you can change the number of "voices" used (the num. of diff. processes), or the distribution of drift rate, starting tone , starting time, and length for each voice. Play around with this parameter and try to see if you can come up with anything reasonable. I couldn't!
The script also generate a figure representing the resulting voices (still thanks to Ken library), that may look like this:
As usual, you can download the file from here.