Abstract—Different types of noise cancellation techniques are prevalent in recent literatures. The performance of a particular technique depends on mean, variance and maximum amplitude of error. At the same time the process time of signal and complexity of practical implementation of circuits is also a measuring tool for performance of a technique. The objective of this paper is to compare performance among the short time Fourier transform (STFT), wavelet transform (WT), least mean square (LMS) and recursive least square (RLS) methods in cancellation of noise from a speech signal. The analysis of the paper provides us the way of selection of the best denoising technique based on the statistical parameters of the above four mentioned techniques.
Index Terms—Signal denoising, CPU time, statistical parameters of adaptive filter, wavelet transform, short-time Fourier transform, least mean square and recursive least square.
Mahbubul Alam and Md. Imdadul Islam are with the Computer Science and Engineering Department, Jahangirnagar University, Savar, Dhaka 1342, Bngladesh. M. R. Amin is with the Department of Electronics and Communications Engineering, East West University, 43 Mohakhali, Dhaka1212, Bangladesh corresponding author, phone: +880-1715296735; fax:+880-2-8812336; (e-mail: ramin@ ewubd.edu).
Cite: Mahbubul Alam, Md. Imdadul Islam, and M. R. Amin, "Performance Comparison of STFT, WT, LMS and RLS Adaptive Algorithms in Denoising of Speech Signal," International Journal of Engineering and Technology vol. 3, no. 3, pp. 235-238, 2011.