My friend did something similar to the concepts in this article, except he applied it to compression of audio. It was basically finding patterns in the audio and transforming these into equations. Interesting stuff.