Research On Speech Enhancement Algorithm Based On Complete Aggregate Empirical Mode Decomposition

The voice signal is often interfered by various noises,which seriously affects the operation of the voice processing system.The speech enhancement technology is a method of suppressing the noise signal as much as possible in the noisy speech signal and improving the speech quality.The main contents of this thesis are as follows:1.A speech enhancement technique based on complete aggregation empirical mode decomposition and wavelet threshold noise reduction is proposed.After the noisy speech has been completely decomposed by empirical mode decomposition,the main components of the speech signal are extracted through the autocorrelation function.The unbiased risk estimation threshold noise reduction method is used to process the speech signal components,and the wavelet soft threshold method is used to process the noise dominant components.Finally reconstruct the speech signal to complete the enhancement.Subsequent experimental comparisons show the superiority of this method.Under the influence of white Gaussian input noise,the signal-to-noise ratio can be improved.2.A speech enhancement technique based on complete aggregation empirical mode decomposition and auditory masking noise reduction method is proposed.After the noisy speech is completely decomposed by empirical mode decomposition,the main components of the speech signal are extracted through mutual information entropy,and the noise power spectrum estimation is completed in combination with the multi-window spectrum estimation method.Finally,the auditory mask noise reduction model is used for noise reduction processing.Reconstruct the speech signal.The experimental comparison shows that the method is effective and the performance is superior when the speech signal is affected by noisy noise.3.Finally,this paper uses MATLAB software to write the graphical user interface(GUI)to realize the speech signal enhancement processing software based on the complete aggregation empirical mode decomposition,which is simple and easy to use.The software has the functions of parameter configuration and speech enhancement processing for the speech data signal imported by the user.