Multi-classification of audio signal based on modified SVM

As one of the important multimedia information carrier, audio signal effectively enriches and satisfies people´s apperception and acquirement of the information; in order to improve the accuracy of audio classification, we adopt the modified SVM that is based on hierarchical clustering analysis and binary decision tree to classify the seven types of audio signal in this paper, a number of the samples are used for training of each audio signal so as to obtain the excellent training templates, and then to test the audio signal. Experimental results show that the method has a good classification performance, compared with the traditional one-to-one and other algorithms, our algorithm not only reduces the training and testing time, but also further improves the accuracy rate, up to over 90%.