I spent the last days studying machine learning and how to train a computer to make things (in this case an image) based on a training set full of midi and audio data from the songs I am currently composing. At first the results were quite messy, unorganized and I didn’t understood why, I let the computer do whatever it thought make more sense and while some interesting iterations occurred, it was not what I was looking for. So I set a unique rule, using numbers corresponding to letters in the English alphabet (for instance 1 for a, 2 for b and so on) and writing one word as a “master rule” I was able to get some very interesting outcomes. This was my favorite iteration of all. More music coming soon.