Vehicle Speech Enhancement Algorithm Based on TanhDBN
Long Qian, Fujian Zheng, Xin Yu Guo, Yingxiang Zuo, Wei Zhou · 2020 IEEE 3rd International Conference of Safe Production and Informatization (IICSPI) · 2020
Vehicle speech enhancement has long been a challenging issue owing to the system that exhibits nonlinearity, strong noise and strong interference in the vehicle environment. In this paper, a vehicle speech enhancement algorithm based on the combination of deep belief network and wiener filtering is proposed. First, the quantum particle swarm optimization (QPSO) algorithm is employed to optimize the parameters of the deep belief networks (DBN) model. Then, to learn better features of the input vehicle Speech signal, a hyperbolic tangent (tanh) function is used to replace the traditional sigmoid function of DBN. Finally, the learned high-level speech feature signals are inputted to the wiener filtering algorithm for speech enhancement. Experimental results show that the proposed method can effectively eliminate the noise of the original speech signal and effectively enhance the processed speech signal. Meanwhile, the information of the original speech signal can be retained.