Research On Dual-Channel Speech Enhancement Technology Based On Differential Microphone Arrays

With the development of miniaturization of electronic devices,some conventional microphone arrays speech enhancement methods limited by the size of the arrays,the number of microphones,and the speed of operation are not suitable for it.The differential microphone arrays are small in size and have superior directionality.The balance between algorithm performance and applicability has been achieved and has received extensive attention and application.In this context,this dissertation studies the existing dual-microphone speech enhancement based on differential microphone arrays and makes some improvements to the algorithms.The details are as follow.First,research and analyze the dual-channel speech enhancement method based on First-order Differential Microphone(FDM)arrays are performed.It is found that there is a problem of residual directional noise in the enhanced speech signal processed by this algorithm,and the speech quality will be degraded when the silent segment of the test speech is not accurately estimated.Secondly,in order to the defect of residual directional noise in the enhanced speech signal processed by the FDM-SS algorithm,a variable step size Least Mean Square(LMS)speech enhancement algorithm based on differential microphone arrays is proposed.The algorithm first performs first-order differential microphone array processing on the noisy speech signals collected by the two microphones,and combines the obtained speech channel signals and noise channel signals as the desired signal and error signal with the variable-step frequency domain LMS algorithm.Control the weight coefficient of the adaptive filter to adjust.The residual directional noise in the speech channel signal is further eliminated.The simulation experiment results show that the improved algorithm proposed can effectively suppress the directional interference noise and improve the speech quality.Finally,in order to the defect that the BM module in the Generalized Sidelobe Cancellation(GSC)structure will lose the speech signal in the low frequency band.Combining the advantages of the first-order differential microphone array,an improved generalized sidelobe cancellation algorithm based on the first-order differential microphone array is proposed.The algorithm uses the speech channel signal and the noise channel signal processed by the differential microphone array as the Fixed Beamformer(FBF)module and the Blocking Matrix(BM)module input of the GSC respectively.And add a BM module to compensate the energy loss of the low frequency part through the equalization filter,and improve the performance of the Adaptive Noise Canceller(ANC).The Signal-to-Interference Ratio(SIR)is estimated by using the two channel signals to control the ANC adaptive filter weight coefficient to update and eliminate the residual noise in the speech channel signal.The experimental results show that the proposed algorithm can effectively overcome the directional noise and reduce the distortion of the speech signal.The objective evaluation of speech quality and the noise reduction effect are better than the reference algorithm.