Comparative Analysis of EMD and VMD Algorithm in Speech Enhancement

• Published in: International journal of natural computing research Vol. 6; no. 1; pp. 17 - 35
• Main Authors: Ram, Rashmirekha, Mohanty, Mihir Narayan
• Format: Journal Article
• Language: English
• Published: Hershey IGI Global 01.01.2017
• Subjects: Algorithms; Communication; Comparative analysis; Data mining; Empirical analysis; Fourier transforms; Mean square errors; Methods; Noise; Noise levels; Security aspects; Speech; Speech processing; Support vector machines; Wavelet transforms; Wiener filtering
• ISSN: 1947-928X; 1947-9298
• DOI: 10.4018/IJNCR.2017010102

Signal enhancement is useful in many areas like social, medicine and engineering. It can be utilized in data mining approach for social and security aspects. Signal decomposition method is an alternative choice due to the elimination of noise and signal enhancement. In this paper, two different algorithms such as Empirical Mode Decomposition (EMD) and Variational Mode Decomposition (VMD) are used. The bands are updated concurrently and adaptively in each mode. That performs better than the traditional methods for non-recursive signals. Further it has been investigated that VMD outperforms EMD due to its self-optimization methods as well as adaptively using Wiener filter. It is shown in the result section. Different noise levels as 0dB, 5dB, 10dB and 15dB are considered for input signal.