Research On Broadband Beamforming Algorithm Based On Microphone Array

In recent years,microphone array technology has been widely used in many fields and has set off a new wave of research.The microphone array-based beamforming technology plays an important role in the high performance of microphone array systems.For broadband signals,the signals received by microphone arrays are frequency dependent and the time delays of different array elements cannot be simply equated to phase shifts,thus making the processing of broadband signals more complex compared to narrowband signals.In this thesis,the broadband beamforming algorithm based on microphone array is studied from the perspective of broadband signals.The main research of this thesis is as follows.1.The improved array optimization algorithm based on narrowband signal is improved,and the improved algorithm divides the broadband signal into frequencies,and takes the average value of the maximum side flap level and the beam direction map at all frequencies as the objective function,which solves the limitation of the existing array optimization algorithm based on narrowband signal,and extends the array optimization algorithm based on narrowband signal to broadband signal,so that the maximum side flap level is the lowest and the gate flap effect is avoided.The maximum bypass level is minimized to avoid the gate effect.The linear array is sparse and a sparse semicircular microphone array model for wide and narrowband signals is proposed to enrich the current microphone array optimization model.The beam shape and width of broadband signals based on microphone array vary according to the frequency.Therefore,the objective function in the optimization of microphone array based on broadband signals is further improved by combining the fixed beam formation directional map function to construct a new objective function so that the beam formation has a constant beam width at different frequencies.2.The direction of arrival(Direction Of Arrival,DOA)estimation algorithm based on broadband signal is improved.The algorithm selects the frequency points with higher energy in the broadband signal,uses the information at these frequency points for DOA estimation,and counts the obtained angle information to obtain the broadband DOA estimation results.The algorithm solves the problem of underutilization of the data received by the microphone array in the existing wideband DOA estimation algorithm and improves the resolution of the wideband DOA estimation.The DOA estimation results are influenced by the signal-to-noise ratio.To further improve the accuracy of the wideband DOA estimation results,the algorithm is further improved by setting a threshold value using the angular accuracy to further filter the obtained angular information to reduce the influence of signal-to-noise ratio and improve the accuracy of DOA estimation.3.The beamforming of single and dual microphone arrays is improved.The singlemicrophone array beamforming is angularly broadened by setting the broadening angle according to the variance of DOA estimation,the main flap of the broadening direction map and the zero trap,which solves the problem of inaccurate beamforming due to the pointing error of DOA estimation.The beamforming algorithm of the dual-microphone array is improved by using the generalized cross-correlation(Generalized Crosscorrelation,GCC)of the outputs of the two single-microphone arrays to calculate the time delay,so as to perform time synchronization and obtain the beam output of the dualmicrophone array after synchronization,which solves the non-synchronization problem in the dual-microphone array.