Source-location-informed sound field recording and reproduction with spherical arrays

Abstract

A sound field recording and reproduction method for spherical arrays that takes into account prior information on the locations of sound sources to be reproduced, i.e., primary sources, is proposed. Current methods using spherical arrays of microphones and loudspeakers, such as higher order ambisonics (HOA), are based on sound field analysis and synthesis in a spherical harmonic domain. However, the size of the reproduced region is limited by the maximum available order of the spherical harmonics, which is usually determined by the microphone array design. Our proposed method exploits approximate positions of primary sources, which may be estimated from the signals received by microphones or other sensor data, to obtain the driving signals of the loudspeakers. Spatial basis functions and their coefficients representing the driving signals are optimized on the basis of maximum a posteriori (MAP) estimation using the prior source locations. Numerical simulation results indicated that the reproduced region can be enlarged to greater than that when using HOA by using the proposed method.