A study of features for speech/music discrimination

摘要

Speech/Music discrimination has been studied for various applications such as automatic indexing of audio data. In this paper, we focus on four acoustic features examined in related studies and evaluate these features with audio data sets classified into seven audio classes.