I did a little bit of machine learning with neural network when I was in school, and I'd like to try it on sounds. What I'd like to do is to identify patterns, types of sounds, like "people talking", "loud, compressed rock music", etc.