Efficiency of Oversampled Discrete Fourier Transform Filter Blocks to Cancel Back- ground Noise Hamed SHIRZAD 1 Javad POORAHMADAZAR 2 Sasan Ahdi REZAIEH 3 1,2,3 Islamic Azad University of Urmia Branch, Behesti, 57153, Urmia, West Azarbayjan Abstract: Slow focus ability and complex computations are the main barrages facing the usage of adaptive noise filtering for cancellation of the background noise. Here we have developed noise canceller by using two fold over sampled filter banks. We had formulated the system by few realistic assumptions to analyses the filter. System offers a structure without cross-filters or gap filter banks and hence decreases the residual noise at the output. Increasing initial convergence rate is addressed and computational complexity is analyzed .The performance under white and colored environments, is evaluated in terms of mean square error performance. As a result fast initial convergence was resulted. An increase in the amount of noise reduction by approximately 5dB compared to full-band model reached under actual speech and background noise. In spite of the insertion of analysis/synthesis filter banks, the proposed noise canceller is still computationally efficient. Key words: Background Noise cancellation, Adaptive Noise Canceling, Gap Filter, DFT Filter. Internatıonal Journal of Natural and Engineering Sciences 4 (1):39-48, 2010 ISSN: 1307-1149, E-ISSN: 2146-0086, www.nobel.gen.tr Corresponding author E-mail: sasan.ahdi.rezaieh@gmail.com INTRODUCTION Noise can seriously damage speech communication especially in noisy environments like crowded streets, factories, noisy rooms. In this article, using the least mean square (LMS) algorithm of adaptive noise filtering and its variants are often used to adapt a full-band filter with a relatively low computation complexity and best performance. However, the full-band LMS solution suffers from significantly degraded performance with colored interfering signals due to the eigen-value spread of the autocorrelation matrix [1]. Moreover, as the length of the adaptive filter is increased, the convergence rate of the LMS algorithm decreases and the computational complexity increases. This can be a problem in applications such as acoustic noise and echo cancellation that demand long adaptive filters to model the path response. These issues are especially important in hand free communication, where processing power must be kept minimum [2]. Sub-band adaptive filtering using multi rate filter banks has been proposed in recent years to speed up the convergence of the (LMS) algorithm and to reduce the computational burden [3],[4]. In this approach, multi rate filter banks are used to split the input signal into a number of frequency bands, each serving as an input to a separate adaptive filter. The sub-band decomposition greatly reduces the update rate of the adaptive filters, resulting in a much lower computational complexity. Therefore, sub-band signals are often down sampled in a sub-band adaptive filter system, this leads to a whitening effect of the input signals and hence an improved convergence behavior [5]. In critically sampled filter banks, the presence of aliasing distortions, requires the use of adaptive cross filters between sub- bands. However systems with cross adaptive filters generally converge slowly and have high computational cost, while gap filter banks produce spectral holes which in turn lead to significant signal distortion [6].In recent literature, the issue of using filter-banks to improve the performance of adaptive filtering is often considered from the view point of application to line echo cancellation in telecommunication systems [7],[8],[9] and [10]. In this paper, an improved sub-band noise cancellation system is derived from an existing full-band model, and then the application to the cancellation of background noise is considered. Few assumptions were made in formulating the system equation and deriving the optimum prototype filter. The proposed oversampled scheme offers a simplified structure that without employing cross-filters or gap filter banks reduces the aliasing level in the sub- bands, and hence decreases the residual noise at the system output. The issue of increasing initial convergence rate Received: December 10, 2009 Accepted: January 26, 2010