An Empirical Study of Feature Extraction Methods for Audio Classification