Research On Sound Source Localization Algorithm Based On Small Microphone Array

The sound source localization technology based on the microphone array involves many technical fields such as acoustics,electronics and array signal processing.With the increasing application of this technology in military and civilian fields,the miniaturization of microphone arrays has gradually become an important research direction.The localization accuracy of a microphone array is usually positively related to its aperture and the number of elements.Many sound source positioning algorithms that can achieve good results in large microphone array have poor performance with small microphone array.In order to improve the localization performance of small microphone array,this paper compares the mainstream sound source localization algorithms in different positioning tasks,different source conditions and different sound field environments.On this basis,three improved sound source localization algorithms are proposed for the practical needs of the task.The main research contents and results are as follows:(1)An improved narrowband sound source localization algorithm for small microphone array is proposed.The principles and specific implementation methods of the mainstream algorithms in narrowband signals are analyzed: minimum variance distortionless response beamforming algorithm,multiple signal classification algorithm,l-Singular Value Decomposition algorithm and orthogonal matching pursuit algorithm.In order to solve the problems of high calculation cost and poor localization accuracy in low SNR of l-algorithm,a new compressive sensing narrowband sound source localization algorithm called SP-SVD algorithm is proposed by combining SVD with subspace pursuit algorithm.The simulation results show that the SP-SVD algorithm has the best comprehensive localization performance under the condition of six-element linear small microphone array narrowband signal model,and the computational efficiency is significantly better than the l-algorithm.(2)An improved broadband sound source location algorithm for small microphone array is proposed.The principle and specific implementation method of traditional sound source localization algorithm extending from narrowband signal to wideband signal are studied.The principle of l-algorithm based on joint subbands is studied.The DCS-SP-SVD algorithm is proposed by combining the distributed compressive sensing model with the SP-SVD algorithm,combining the contribution values of the observation vectors in multiple subbands to select the atoms to join the observation set.The simulation results show that the DCS-SP-SVD algorithm has the best comprehensive positioning performance under the condition of six-element linear small microphone array broadband signal model,and the computational efficiency is significantly better than thel-algorithm based on joint subbands.(3)An improved 3D sound source localization algorithm for small microphone array is proposed.Through simulation experiments and theoretical analysis,it is concluded that under the condition of 4 array elements,the Time Difference of Arrivallike algorithm can achieve the optimal localization performance with a regular tetrahedron structure;The principle and implementation of traditional TDOA algorithm and Steered Response Power-Phase Transform algorithm are studied.Aiming at the high computational cost of SRP-PHAT algorithm,BSO-SRP-PHAT algorithm is proposed by introducing Beetle Swarm Optimization algorithm to replace the traversal search process in the original algorithm.Simulation results show that the improved algorithm greatly improves the computational efficiency without compromising the performance of SRP-PHAT algorithm.A small microphone array sound source localization hardware system with regular tetrahedral structure is constructed,and the localization performance of the system is verified by experiments.