Sound field interpolation via sparse plane wave decomposition for 6DoF immersive audio

Volumetric capture of audio for six-degrees-of-freedom (6DoF) reproduction requires recording a sound field at multiple positions using microphone arrays. When microphone arrays capable of recording higher-order Ambisonics (HOA) are used, rendering of 6DoF audio becomes an interpolation problem that involves the calculation of HOA signals at a location intermediate to the original recording positions. We present a sound field interpolation method using multi-point sparse-plane decomposition followed by directional interpolation. The proposed method operates in the time-frequency domain and relies on the decomposition of time-frequency bins into a dominant directional component and a residual which are interpolated separately. The directional component which represents the dominant sound source is interpolated by translation of the associated plane wave components calculated for each microphone array to the interpolation position and calculation of a single, interpolated plane wave. We present an objective validation of the method based on the directional statistics of the interpolated sound field.