Low-SNR Speech Enhancement and Separation in Driving Environment

Noise is the key factor related to the voice recognition performance, and there are complex noises in driving environments. In this paper, the low frequency suppression pretreatment of mixed signal with noise is proposed; Secondly, the first-order recursive smoothing noise estimation algorithm is improved, and bias compensation algorithm is added to reduce the estimation error. Furthermore, the FastICA algorithm based on negative entropy is improved to reduce the impact of initial value sensitivity. As verified by experiments, the proposed algorithms have improved the target driver's speech quality, and much suitable to the speech enhancement and separation in the low-SNR driving environment.