Unsupervised speech/music classification using one-class support vector machines

Audio classification is an important issue in current audio processing and content analysis researches. Speech/music classification is one of the most interesting branches of audio signal classification. In this paper we present an unsupervised clustering method, based on one-class support vector machines (OCSVM) and inspired by the classical K-means algorithm, which effectively classifies speech/music signals. First, relevant features are extracted from audio files. Then in an iterative K-means like algorithm, after initializing centers, each cluster is refined using a one-class support vector machine. The experimental results show that the clustering method, which can be easily implemented, performs better than other methods implemented on the same database.