Open AccessResearch on Speech Signal Denoising Algorithm Based on Wavelet Analysis
Author(s) -
Tan Liu,
Yehui Chen,
Fengzhi Wu
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1627/1/012027
Subject(s) - noise reduction , white noise , step detection , wavelet , additive white gaussian noise , video denoising , noise (video) , gaussian noise , computer science , speech recognition , signal (programming language) , pattern recognition (psychology) , algorithm , artificial intelligence , mathematics , computer vision , telecommunications , filter (signal processing) , object (grammar) , video tracking , multiview video coding , image (mathematics) , programming language
In this paper, wavelet analysis is used to remove noise superimposed on the speech signal to achieve denoising of the speech signal. The prepared speech signal is superimposed with Gaussian white noise with different signal-to-noise ratios, and is removed by using a forced noise removal processing method and a given threshold denoising processing method; The wavelet soft and hard threshold denoising method is used to separately superimpose different Gaussian white noise speech signals, and the hard threshold and soft threshold denoising are respectively performed, and different wavelets are used for processing; Four threshold selection methods are used to denoise the noisy signal using different wavelets. The denoising effects of these methods are simulated in MATLAB software, and combined with the signal-to-noise ratio of the signal after denoising, the denoising effects of various methods are judged. Experimental results show that soft threshold denoising is slightly better than hard threshold denoising. Among the four threshold selection methods, Rigrsure rule denoising is the best. Although other methods are effective, they are not obvious.