Noise Reduction of Speech Signal Based on Wavelet Transform with Improved Threshold Function

The wavelet threshold noise reduction method is widely used in the field of signal noise reduction. Based on the traditional soft and hard threshold function noise reduction, combined with the shortcomings of the current improved wavelet threshold noise reduction algorithm, this paper proposes a new improved speech noise reduction threshold function. The experimental simulation waveform and quantitative indicators show that the threshold function retains effective speech information to a greater extent, and has a better improvement on the continuity and constant error in the process of speech noise reduction. It is suitable for the analysis and processing of noise-containing signals, and has certain practical value.