Preliminary Study On The Acoustic Source Positioning System Based On Microphone Arrays

Sound source localization system based on microphone arrays is a kind of technology that locates sound source by using acoustic and electronic devices receiving and processing the signals of the sound field, which has been widely used in military, industrial, and civilian other areas.The key to the technology of sound source localization based on sound arrival time difference is the estimation of time delay values, algorithm of which is various. In this paper, the author studies the time delay estimation algorithm with generalized cross-correlation, and applies the algorithm in the actual environment. In the design of microphone array models, we focus on the position theory and algorithms of space four-element array model and space five-element array model. Combined with the virtual instrument and Matlab, sound source localization system based on microphone array is developed, which is used to achieve a single sound source localization experiments in real environment. This experiment, which is tested by moving the sound source along the x and y directions, is on the theoretical basis of positioning technology of sound arrival time difference. The results show that the positioning system has a higher accuracy when the distance between the target sound source and the microphone arrays surface is within1.8m in y axis. When more than1.8m, there will be some deviation. When the point of the sound source moves along the x direction with in the range of-0.2and0.3, the positioning result is very accurate; instead, the positioning result has certain volatility. In addition,the stability of the positioning results for the five-element array model is better than the four-element array model’s.Finally, the paper initially investigates multiple sound sources positioning system that depends on sound source separation technology. Combined with LabView, simulation experiment for multi sound source separation is earned out, the theoretical basis of which is the fast Independent Component Analysis algorithm of negentropy. On this basis, two independent source signals have been simulated, and they form two mixed signals after mixed, accurate separation of the mixed signal is verified by simulation. The experiment verifies the feasibility of the separation algorithm, and it builds up the foundation of future multi-sound source localization and visualization.