Wavelet Analysis for Denoising Emotional Speech Signals

  • Anuja, Sanjeev Kumar Dhull


Demand of affective computing is growing rapidly for different applications. Emotional speech signals need efficient and effective denoising techniques and computation to obtain meaningful results. Speech signals are subdivided by using discrete wavelet transform’s mother wavelet functions followed by thresholding. Four different mother wavelets i.e Haar, db4, bior3.5, sym4 tested four performance parameters i.e MSE, SNR, PSNR, CC for different speech emotion signals .The derived results compared to get highest value of SNR,PSNR  and lowest value of MSE among the wavelets applied. In the comparison db4 mother wavelet has performed over other wavelets on the basis of MSE, SNR, PSNR, CC. Matlab and wavelet toolbox have been used for implementation of algorithms. On comparing db4 wavelet gives good SNR, PSNR, CC value after denoising.

How to Cite
Anuja, Sanjeev Kumar Dhull. (2020). Wavelet Analysis for Denoising Emotional Speech Signals. International Journal of Advanced Science and Technology, 29(06), 4862 - 4871. Retrieved from https://sersc.org/journals/index.php/IJAST/article/view/19412