Directed Acyclic Graphs for Content Based Sound, Musical Genre, and Speech Emotion Classification