Stereo Audio Rendering for Personal Sound Zones Using a Binaural Spatially Adaptive Neural Network (BSANN)

A binaural rendering framework for personal sound zones (PSZs) is proposed to enable multiple listeners to receive independent stereo audio programs. Current PSZ systems typically rely on monophonic rendering and therefore cannot control the left and right ears separately, which limits the quality and precision of spatial imaging. The proposed method employs a Binaural Spatially Adaptive Neural Network (BSANN) to generate ear-optimized loudspeaker filters conditioned on the listeners’ positions. The framework integrates anechoically measured loudspeaker frequency responses, analytically modeled transducer directivity, and rigid-sphere head-related transfer functions (HRTFs) to improve the physical realism of the training acoustic transfer functions (ATFs) and spatial rendering fidelity. An explicit active crosstalk cancellation (XTC) stage further improves interaural separation for loudspeaker-based binaural reproduction. Physical measurements in two real rooms, with additional static listener configurations tested in one of them, show that the proposed BSANN framework achieves inter-zone isolation (IZI), inter-program isolation (IPI), and XTC of approximately 9–11 dB for a representative two-listener arrangement without room-specific retraining or calibration. The same model maintains all reported metrics above approximately 7.7 dB across the additional static listener configurations, and above 8.6 dB in a second room whose reverberation time exceeds the training reverberation range. A complementary held-out simulated-room evaluation across 110 rooms spanning $RT_{60}=0.10$–0.60 s shows gradual degradation as the direct-to-reverberant ratio (D/R) decreases, while the objective metrics remain positive throughout the tested range, thereby characterizing the sensitivity of calibration-free binaural PSZ rendering to increasingly reverberant conditions.