Music classification is a core task in the field of Music Information Retrieval (MIR). Classification refers to recognizing patterns in data. Music classification assigns genre, style, mood and etc. to each piece of music, to facilitate managing music data. It is an interesting topic in MIR with potential applications. There has been a considerable deal of attention focused on variety issues of music classification, such as selection appropriate feature sets, feature selection techniques, classification algorithm, etc. In this thesis, a series of empirical experiments are conducted to investigate and evaluate the genre and style classification in music. To validate our investigations and evaluations, several methods are proposed to analyze and interpret the results. In addition, we also design and implement an effective classification approach that obtains higher classification accuracy.
展开▼
