These little critter are generated based on the audio. The style of the body, legs, eyes, eyebrows, mouth etc is randomized (otherwise it would get pretty boring really fast) but the size etc is set by the audio.

They dance and sing based on the beat detection included in minim.