Localization Based Beamforming Approach to Audio Source Separation from Binaural Mixtures

Abstract

This paper presents an approach of localization on the basis of audio source separation from binaural mixtures with adaptive beamforming. A new source localization technique is proposed to spatially localize the sources with two degrees of freedom (azimuth and elevation). We used a head related transfer function to synthesize the binaural mixture that introduces the interaural differences to the source signals. The interaural cues are technically employed to estimate the directions of arrival of the sound sources. Data independent minimum variance beamfilter is used to segregate the sources sequentially to overcome the underdetermined condition (more sources than sensors). We have presented the separation performance by human hearing experiment comparing the original and separated signals. The experimental results confirm the improved performance and show a noticeable efficiency in audio source separation using the proposed source localization technique.