Performance Of Empirical Mode Decomposition And Wavelet Transform In Denoising Of Audio Signal
Audio signal denoising is one of the major technique in multimedia applications. It has long been a topic of
research, yet there is a need for improvement always. There are so many ways to enhance the signal quality or to reconstruct
the signal. In this paper two methods namely Empirical Mode Decomposition (EMD) and Discrete Wavelet Transform
(DWT) are used for reconstruction/denoising of audio signals which are corrupted by additive white Gaussian noise. In
DWT Thresholding is applied to wavelet coefficients of noise corrupted signal, to restore the smoothness of original signal.
EMD decomposes the noisy signal into amplitude and frequency modulated signals called Intrinsic mode functions(IMF’s) .
Hard Thresholding of IMF’s are performed to remove the noise. Performance of both the methods are measured by signal to
noise ratio(SNR) and mean squared error(MSE).
Keywords— Signal denoising, Empirical mode decomposition, Wavelets, Thresholding, Noise reduction.