Wavelet transform and variable-step least mean square algorithm-based voice denoising method

Abstract

The invention relates to a wavelet transform and variable-step least mean square algorithm-based voice denoising method. The method is technically characterized by comprising the following steps of: reducing the dispersion degree of an input vector self-correlation matrix characteristic value of an adaptive filter by using a time-frequency local characteristic of a wavelet, and increasing step factors of the algorithm; and simultaneously, establishing a non-linear function relationship between the step factors and an error signal to ensure that the step factors are adaptively increased at the initial stage and the time change stage and are adaptively decreased at the steady state stage. The method not only can ensure a relatively high convergence speed and little maladjustment but also has certain robustness and denoising performance, so a better denoising effect can be obtained by combining the wavelet transform and the variable-step least mean square algorithm.

Description

Technical field [0001] The invention relates to a speech noise reduction method based on wavelet transform and a variable step size minimum mean square algorithm, which can be applied to various speech signal noise reduction processing systems. Background technique [0002] In the process of voice communication, it will inevitably be interfered by noise introduced from the surrounding environment and transmission media, electrical noise inside the communication device, and even other speakers. These interferences ultimately make the voice received by the listener no longer the original pure voice signal, but the noise-contaminated voice signal. As an effective means of controlling low-frequency noise, adaptive voice active noise reduction technology has been greatly developed in recent years. [0003] The adaptive noise cancellation system is a typical application of adaptive voice active noise reduction technology. It is an adaptive filter with two inputs: the adaptive filter run...

AI Technical Summary

Problems solved by technology

Although researchers at home and abroad have done a lot of work on adaptive filtering algorithms and put forward some improved algorithms, there are still many problems: (1) The contradiction between convergence speed and steady-state error cannot be fundamentally overcome: step If the length factor is large, the convergence speed will be fast, but the misalignment will be large; if the step size factor is small, the misalignment will be small, but the convergence speed will be slow; (2) The algorithm is sensitive to noise, which is more obvious only in environments with high signal-to-noise ratios. (3) The convergence speed is sensitive to the distribution of the eigenvalues ​​of the autocorrelation function matrix of the input signal: if the distribution is too scattered, the maximum If the difference from the minimum value is too large, the convergence speed will be very slow

Therefore, directly applying the existing adaptive filtering algorithm to the noise reduction system will cause problems such as slow convergence speed, large steady-state error, and low computational efficiency.

Images