Research On The Microphone Array Speech Enhancement Algorithms In Driving Environment

Intelligent electronic control systems,such as portable navigation systems,smart rearview mirrors and entertainment systems are widely used in cars with the development of technology.Most of them achieve their operational functions through the buttons,switches and touch panel at present.These manual operations may distract the attention of drivers and produce security risks.Therefore,man-machine interaction by acoustic control technology is one of the important research directions in recent years.However,driving environment is a complex acoustic environment which is filled with various noises.In such environment,the technology of speech recognition at present is difficult to meet our needs.As a consequence,research on technology of speech enhancement in situation of driving is significant for future design and development of automotive electronic products.The microphone array speech enhancement algorithm for driving environment is studied in this paper.The main work is as follows:The characteristics of the speech signal and automobile noise are analyzed in this paper.According to the characteristics of automobile noise,low frequency suppression is put forward.Amplitude spectrum cutting is introduced to correct power spectrum,which improves the relative signal-to-noise ratio between speech existence and speech inexistence,then the speech existence part becomes more prominent and generalized cross-correlation is used to compute time delay between different channels.The proposed method improves the accuracy of time delay estimation.In order to deal with driving environment,generalized sidelobe canceller is improved combined with the signal propagation model in near field.Blocking matrix is ameliorated with source location information,and amplitude compensation is introduced by adjacent channels in the noise estimation process,then the logistic function is utilized to control the noise gain.The improved algorithm can effectively reduce the leakage of the target speech signal and improve the speech quality.An optimization scheme is proposed to improve the shortage of generalized sidelobe canceller in non-coherent noise suppression.Multi-channel spectral subtraction algorithm is adopted to deal with the signal which is enhanced by the generalized sidelobe canceller.Single channel speech enhancement algorithm and multi-channel speech enhancement algorithm are combined to promote the noise suppression ability.Finally,efficient voice activity detection algorithm using long-term speech information is adopted to extract the target speech.The results of the experiment show that the proposed algorithm can be applied to the speech enhancement in driving environment,and it has superior improvement on SNR and speech quality compared with the existing algorithms.