The Research On Algorithms Of Indoor Sound Source Localization Based On Microphone Array

With the development of multimedia technology, the array signal processing technology has been widely used in lots of areas. The technology of sound source localization based on microphone array is one of the key techniques in array signal processing and a new research field of speech signal processing. This paper using indoor speech positioning technology based on microphone array, mainly done the following work:1、Based on the deep research of the sound source localization of microphone array, this thesis has analyzed the influence factors and the difficulties, then established the microphone array speech model and given the methods of preprocessing on speech signal.2、On the basis of analyzing the low precision and poor real-time positioning of traditional method, we proposed a sound source localization algorithm based on 7-microphone array and PHAT-GCC(phase transform based generalized cross correlation function) method. In this proposed algorithm, seven-microphone array is established to receive the sound source signals, the signals are preprocessed via spectral subtraction and cepstrum technology to obtain the pure speech signals, and the TDOA(time difference of arrival) method are used to calculate the azimuth angle, pitching angle, and distance. At last the sound source localization is determined by using the geometric method with angle and distance. And comparison with the conventional four-element cross array.3、In view of the problems of inaccurate localization and ambiguous orientation of traditional Cross power Spectrum Phase(CSP) in direction of arrival estimation, an Circular Integrated Cross power Spectrum(CICS) algorithm of sound source localization using microphone array was proposed. In this proposed algorithm, twelve element circular microphone array is designed, the phase rotation factor is defined by combining the time delay of speech signals with the phase information of speech signals received by each pair of microphone and introduced into the traditional cross spectrum to define novel Circular Integrated Cross power Spectrum, the sound source orientation can be estimated by CICS algorithm and can improve the estimation performance of direction of arrival.4、In view of the existing shortcomings of the study on multiple sound source localization method, a dual sound source localization algorithm based on consistent focusing transformation least squares using microphone array was proposed. In the proposed algorithm, six element circular microphone array is designed. To the received speech signal, the signal covariance matrix is obtained and the center frequency bin is defined in the range of frequency. Then the focusing transformation matrix is defined according to the consistent focusing transformation existed of not changing angle in a certain range of any angle, and solved by the least square method. Finally, in the different center frequency bins of bandwidth, the Multiple Signal Classification (MUSIC) method is used to obtain the signal space spectrum through the consistent focusing matrix obtained by the least squares. In the actual condition of only received signal can be taken, the sound sources localization are estimated by using the average signal space frequency mean and time snapshots estimation method and improved the localization accuracy.5、This chapter analyses the above algorithm for sound source localization, and the algorithms of the fourth and the fifth chapters are realized by the sound source localization system. The experiment results show that the proposed sound source location algorithm has high accuracy that meets the actual needs. Finally, summed up the work of this paper and made the prospect of improvement possible.