Separation and mixing parameters estimation for localization in distance based on features extraction

In the current literature a big progress has been achieved in Azimuth localization, however little attention has been paid to localization in distance. In this paper we present a new technique for mixing parameters determination used for the sound source localization in distance using only two observed mixed speech. The mixing parameters are estimated by determining the slope of a scatter plot of the separated sound source from the time delayed mixed speech captured by each microphone. The proposed model is based on Time Delay of Arrival (TDOA) estimation, and sound source separation technique. The separation task is performed by a new approach based on sources features extraction. It relies on harmonics enhancement using sources excitation characteristics. Separation performance was evaluated in terms of objective metrics. The proposed model offers a better separation performance compared to other technique in the field, therefore it promises more accurate results in mixing parameters determination.