An optimal control point distribution method for two-channel crosstalk cancellation

Binaural audio rendering aims to reconstruct virtual sound images at a listener's ears using headphones or loudspeakers. While headphones can deliver stereo sound separately to each ear, loudspeakers-based binaural audio systems suffer from intrinsic crosstalk, which necessitates crosstalk cancellation (CTC) as a preprocessing technique. Traditional two-channel CTC methods experience performance degradation near ill-conditioned frequencies that are determined by physical system configuration. The optimal source distribution approach overcomes this problem by varying the loudspeaker pair's positions with frequency; however, its application is constrained by the requirement for a large number of loudspeakers. As an alternative approach, this paper proposes an optimal control point distribution (OCPD) method that minimizes the condition number of the transfer matrix by dynamically adjusting control point positions instead of the loudspeaker positions. By selecting appropriate control point locations, the proposed OCPD method effectively mitigates the performance deterioration of traditional two-channel loudspeaker systems at the ill-conditioned frequencies without increasing the number of loudspeakers. Theoretical analysis and simulation results demonstrate that the proposed OCPD method outperforms conventional two-channel CTC systems and, using only two loudspeakers, achieves performance comparable to the discrete optimal source distribution method with five loudspeaker pairs. Experiments conducted in an anechoic chamber validate the theoretical findings and further confirm the effectiveness of the proposed approach.