P3-11: Perceptual Comparison of Loudspeaker and Binaural Reproduction Systems for Higher-Order Ambisonic Speech Auralization

Zhang, Yile*, Izumi, Yuto, Trevino, Jorge, Abhayapala, Thushara, Samarasinghe, Prasanga, Bastine, Amy

Subjects (starting with primary): Spatial audio recording and reproduction ; Modelling, analysis and synthesis of acoustic environments

Presented in Poster Session 3

Abstract:

Higher-Order Ambisonics is widely used to render and evaluate sound fields using loudspeaker and binaural reproduction systems, yet the perceptual relationship between these reproduction modes remains an important practical question. This study compared auditory perception across three reproduction systems: a rectangular 42-channel physical loudspeaker array, a binaural rendering using the equivalent 42-channel layout, and a binaural rendering based on a near-spherical 122-channel loudspeaker layout. Measured fourth-order Ambisonic room impulse responses from nine rooms were processed using two Ambisonic decoding methods, namely the All-Round Ambisonic Decoder and the Higher-Order Spatial Impulse Response Rendering. The resulting responses were convolved with anechoic Japanese speech to generate stimuli, which were evaluated by 13 participants. Two attributes were assessed: perceived area of the sound source and speech naturalness. Results showed that perceived source area broadly followed the reverberation characteristics of the rooms, with significant inter-participant rank agreement in all six decoder-reproduction system conditions. In contrast, naturalness ratings showed greater variation across reproduction systems. The physical loudspeaker system showed the strongest inter-participant rank agreement, whereas its 42-channel binaural counterpart showed weak agreement. The 122-channel binaural condition showed higher agreement for naturalness than the 42-channel binaural condition.

This PDF is password protected. Enter the proceedings password in the PDF viewer when prompted.