Research On Sound Source Localization Techno-Logy Based On Digital Microphone Array

Thanks to the rapid development of artificial intelligence technology and the booming voice interaction market,microphone arrays have gradually emerged in recent years and have been widely used in voice recognition and control,vehicle systems and other fields.As a physical interface,the microphone array has limited gain to voice signals,and other functions such as enhancement of voice signals are required to cooperate with it.Sound source localization technology has attracted attention in recent years and attracted a large number of scientific researchers to study it.Traditional sound source localization systems have disadvantages such as large size,poor real-time performance,and poor noise immunity of the sound source localization algorithms used.In order to improve the accuracy of the sound source localization system,deeper and more extensive research is needed.This paper studies the structure of the microphone array,sound source localization algorithms,and speech signal decoding methods based on digital microphone arrays.The specific contents are as follows:(1)The actual collected voice signal contains noise,which will have a great impact on the processing result of the voice signal,so It is necessary to preprocess the voice signal.This article introduces and simulates commonly used voice signal preprocessing methods.Based on conventional beamforming algorithms,the performance of four commonly used microphone arrays is simulated and analyzed.Based on the simulation results,this paper takes the rectangular microphone array as the research object,and derives the sound source localization formula of the four-element rectangular microphone array.(2)In the sound source localization algorithm,based on the generalized quadratic cross-correlation algorithm weighted by the phat function,in order to reduce the influence of noise on the performance of the algorithm,the phat function is improved,and analyzed the time delay estimation accuracy of the generalized quadratic cross-correlation time delay estimation positioning algorithm weighted by the improved phat function under different signal-to-noise ratios.The calculation results show that the improved algorithm has improved stability and noise immunity.After that,the improved generalized quadratic cross-correlation algorithm is combined with the conventional beamforming algorithm,and a beamforming algorithm based on search area division is proposed,and the real-time positioning performance is theoretically analyzed.The result shows that compared with the conventional beamforming algorithm,the calculation time of this algorithm is greatly reduced.(3)Based on FPGA,the synchronous data collection,packing and transmission of the four-element rectangular microphone array have been completed.Due to the voice signal output by the digital microphone used in this article is PDM encoded,this voice signal cannot be played directly by the music player.This article studies the PDM voice signal decoding method based on FPGA,and simulates it in Modelsim to verify the feasibility of its implementation on FPGA.(4)Based on theoretical simulation,a sound source localization system was built and related experiments were carried out.The research results show that the designed improved phat function weighted generalized quadratic cross-correlation time delay estimation positioning algorithm has better noise resistance than the traditional time delay estimation positioning algorithm.The proposed beamforming algorithm for search area division is less computationally consuming,and it can meet the real-time requirements of the sound source positioning system in daily use.