Wavelet-Based feature extraction for musical genre classification using support vector machines

Musical genre classification task falls into two major stages: feature extraction and classification. The latter implies a choice of a variety of machine leaning methods, as support vector machines, neural networks, etc. However, the former stage provides much more creativity in development of musical genre classification system and it plays crucial part in performance of the system as a whole. In this paper we present initial study of waveletbased feature extraction in the task of musical genre classification. A new type of feature vector, based on continuous wavelet transform of input audio data is proposed. The method of feature extraction was tested using support vector machine as a classifier. The results of our experimental study are shown.