Abstract Conventional loudspeaker-based binaural reproduction minimizes the pressure error at each ear while incorporating regularization to control loudspeaker effort. However, this approach can be disadvantageous for spatial audio applications, as it conflates two distinct perceptual objectives, the spectral fidelity and the spatial accuracy, while offering only a single tuning parameter, the regularization factor (λ). This study introduces a cue-constrained Tikhonov (CCT) approach that transforms the pressure matching objective into a set of binaural perceptually driven subobjectives. The formulation is illustrated for a case study where two listeners sit side by side. For two listeners watching the same audio program, the optimization method effectively aims to match the average of the left ear signals at both listeners with the reference signal, enforces similarity between the left ear signals across listeners, and targets the interaural level and time differences within each listener’s head. Simple row normalization eliminates arbitrary scale factors, enabling user-supplied weights to control the priorities for the specified subobjectives. Results suggest that this method is advantageous for cases where spatial accuracy and low drive effort are priorities and some spectral imperfections can be tolerated.
Erspamer et al. (Fri,) studied this question.