Sparsity-based higher order ambisonics reproduction via LASSO

The under-determined problem in Higher Order Ambisonics (HOA) makes it hard for larger accurate reproduction area especially at high frequencies. In this paper, the least absolute shrinkage and selection operator (LASSO) algorithm is used to solve the problem by obtaining the sparse representation of the original plane wave field. The algorithm minimizes the squared error between the original and reproduced microphone signals with constraint on plane wave gains, so as to select a few nonzero plane wave gains for reproduction. Simulation results demonstrate that sparsity-based plane wave field reproduction via LASSO achieves small reproduction errors and outperforms basic decoding. In addition, fewer loudspeakers are active to reproduce the original field and have better performance.