New De-noising Method for Speech Signal Based on Wavelet Entropy and Adaptive Threshold

Xiaojuan Chen

Journal of Information and Computational Science · 2015 · 15 citations · 1 references

Concepts

Abstract

Aiming at the problem of speech signal de-noising, this paper presents an adaptive threshold denoising method based on wavelet entropy. The method decomposes speech signal with noise by wavelet transform, and calculates wavelet entropy of the decomposed signal in each wavelet sub-interval. It combines the wavelet entropy with adaptive threshold to determine the threshold of high frequency coe‐cients. Compromised index threshold function is proposed to denoise the speech signal, and then the denoised signal is reconstructed. Finally, the paper compares the de-noising performance of the proposed threshold method, minimaxi threshold method, sqtwolog threshold method, and rigrsure threshold method. The simulation results show that when the input Signal-to-noise Ratio (SNR) is 7 dB, the output SNR with the proposed threshold de-noising method is the largest, and the input and output SNR curve is higher than what the other three kinds of threshold de-noising methods have, which proves that this method has better de-noising performance.

References

1