Abstract: The sound field separation method has been widely used for recovering the target sound field in the mixed sound field.However,the methods bear heavy measurement load due to the restriction of sampling theorem.In this study,a sound field separation method was proposed based on compressive sensing.The method consists in representing the sound field in a sparse basis by using the singular value decomposition,and utilizing the sparse regularization to find a sparse solution.The method is validated by both numerical and physical experiments.Comparing to the traditional equivalent source method-based separation,the proposed method has the advantages both on the accuracy and the spatial resolution for under-sampled pressure data.