Machine learning can be a hard subject to find a way into. The “two robots” explanation here is a nice analogy. I find that if I can get a conceptual foothold then asking the right questions to move forward becomes easier.

(Also, the Awesome Echo Loopsynth - yes, that’s it’s name - that’s the subject of the linked post is a fabulous thing I backed on Kickstarter. It takes sound from the environment and processes it into interesting new shapes, which is something I’ve been hooked on since the RjDj iPhone app appeared [it’s now called Hear, at least for the moment])