The spectral quality reproduced by multichannel spatial audio reproduction methods are studied in two regular and one irregular loudspeaker layouts when the human head is analytically modeled as a spherical scatterer. The layouts have such a number of loudspeakers that is found in some high-end laboratory conditions. The results verify that the reproduction quality drops above a certain frequency limit, and the introduced spectral errors are more severe with Ambisonics than with VBAP. The head-like scatterer increases errors on the contralateral side of a sound source, and the errors on the ipsilateral side are mostly unaffected. In addition, the frequency limit of the errors may be lower depending on the source direction.
Authors: Lauros Pajunen (Aalto University), Archontis Politis (Tampere University), Markus Vaalgamaa (Huawei Technologies), Stefan Strömmer (Huawei Technologies) and Ville Pulkki (Aalto University)