EVALUATION OF GENERALIZED CROSS CORRELATION PHASE TRANSFORM LOCALIZATION SYSTEM IN REVER-BRANT ENVIRONMENTS WITH A MICROPHONE ARRAY

Abstract

3D sound for virtual and augmented reality is proposed by numerous researchers to provide re-alism in a graphical context. Producing 3D sound poses many challenges, such as personalized head related transfer function for binauralization and the capturing and tracking of users by mi-crophone array or tracking sensors. In this paper, a framework for localization of the user posi-tion in the 3D-space based on generalized cross correlation weighting with phase transform function is implemented. It is then evaluated to achieve the goal of improving speech communi-cation in real noisy environments. The effects of ambient noise on the performance of the local-ization framework in real-time performances were evaluated with different microphone arrays such as linear, circular, and spherical. For the evaluation of the framework, three types of mi-crophone arrays were designed. Each array configuration was designed with a theoretical simu-lation to achieve omni-directional characteristics in the frequency range of human speech. Ex-periments were performed in three different environments with different RT and background noise levels.