Sparse sound field decomposition using group sparse Bayesian learning

A sparse sound field decomposition method is proposed. Sound field decomposition is the foundation of the various acoustic signal processing applications and enables the estimation of the entire sound field from pressure measurements. The plane wave decomposition, i.e., spatial Fourier analysis, of the sound field has been widely used; however, artifacts originating from spatial aliasing occur above the spatial Nyquist frequency. We have proposed a sparse sound field decomposition method based on a generative model as a sum of monopole source and plane wave components in the context of sound field recording and reproduction. For more accurate and robust decomposition, we propose three different group sparse signal models based on physical properties and a decomposition algorithm by extending sparse Bayesian learning. In simulation experiments, the accuracy of sparse decomposition was improved compared with that of current methods.