A Homomorphic Filtering Speech Enhancement Method Based on Broadband Noise

Abstract

The invention discloses a homomorphic filtering speech enhancement method based on wideband noises, and belongs to the speech processing method technology field. The object of the invention is to solve the problems of the prior art such as easy generating of errors during extraction of speech signal characteristics and a lot of residual noises of denoised speech signals. The homomorphic filtering speech enhancement method comprises steps that short-term stationary speech signals are acquired, and the auto-correlation analysis of the speech signals is carried out to acquire auto-correlation coefficients; the homomorphic filtering analyzing processing of the auto-correlation coefficients is carried out, and during the homomorphic filtering analyzing processing, the cepstrums of the speech signals having the noises are acquired; the noise components of the cepstrums of the speech signals having the noises are removed, and cepstrums of enhanced speeches are acquired; characteristic parameters after the denoising are acquired by spectral analysis, and the speech signals after the denoising are synthesized. The improved homomorphic filtering speech enhancement method is used to enhance robustness of speech cepstrum characteristics on ambient noises, and the characteristic information of the speeches are acquired accurately, and therefore the robustness is better, and an object of realizing the speech enhancement is achieved in a better way.

Description

technical field [0001] The invention specifically relates to a homomorphic filtering speech enhancement method based on broadband noise, and belongs to the technical field of speech processing methods. Background technique [0002] In the actual environment, the speech signal processing technology includes many aspects, and the speech enhancement technology is one of the effective methods to solve the noise pollution of the speech. The goal of its research is to extract the pure speech from the noisy speech signal as much as possible in the system output. There are many kinds of traditional speech enhancement techniques, such as spectral subtraction, central filtering method, homomorphic filtering anti-noise method, nonlinear processing method, adaptive analysis method, wavelet analysis method, etc., noise processing process Among them, different speech enhancement methods will be selected according to the different characteristics of speech, the perception characteristics o...

AI Technical Summary

Problems solved by technology

[0009] Therefore, aiming at the problem that the existing technology is still prone to errors when extracting the features of the speech signal, and the speech signal after noise reduction still has a lot of residual noise, the present invention provides a speech enhancement method based on broadband noise homomorphic filtering

Images