点击次数：

论文类型：期刊论文

发表时间：2015-09-20

发表刊物：Journal of Information and Computational Science

收录刊物：EI、Scopus

卷号：12

期号：14

页面范围：5459-5467

ISSN号：15487741

摘要：A novel SVM-based binaural sound source localization is proposed in this paper. The binaural sound signal is first processed by Gammatone filterbank, and the two ITD and ILD cues are extracted based on binaural auditory front-end. Then the cues are concatenated into feather vector and provided to SVM classifier which learns patterns between azimuth and binaural cues, and estimates azimuth of sound source. Simulation results show that the proposed method has robust performance for the sound source localization in different noise and reverberation conditions. Copyright ? 2015 Binary Information Press.