Constrained optimization of acoustic contrast for personal sound zones based on array effort control

Abstract

Personal sound zone (PSZ) systems leverage loudspeaker arrays to deliver individualized audio content to multiple listeners within a shared acoustic space. A fundamental challenge in such systems lies in simultaneously achieving high acoustic contrast between bright and dark zones while maintaining low signal distortion in the listening zone, attributable to constraints in array geometry and the inherent complexity of the sound field. To maximize the utilization of loudspeaker array capability through quantitative trade-offs between acoustic contrast and signal distortion, constrained optimization with acoustic contrast as the primary objective has been explored. However, existing approaches exhibit high sensitivity to fixed regularization parameters, often leading to issues such as acoustic contrast oversaturation or excessive loudspeaker weight amplitudes. To eliminate manual regularization parameter selection—thus improving the practicality and robustness of PSZ design, we propose a novel acoustic contrast-constrained optimization method incorporating adaptive array effort control. By dynamically adjusting regularization parameters, the method achieves a balanced trade-off between acoustic contrast control accuracy and loudspeaker weight amplitude restriction. Furthermore, compared to conventional adaptive regularization techniques based on loudspeaker weight energy constraint, our method demonstrates superior control over array energy gain, thereby preventing the excessive loudspeaker power demand for generating desired bright zone acoustic energy under high-contrast conditions. When the predefined acoustic contrast target exceeds system capability, the proposed method automatically relaxes the requirement to the maximum achievable contrast under the given array effort constraint. Finally, we investigate the effects of array effort control degree and filter length on system performance, evaluating both frequency-domain design performance and time-domain filtering behavior with comparative simulations and experiments.