Improving Speech Signal Intelligibility by Optimal Computation using Single-Channel Adaptive Filtering
Ohidujjaman Ohidujjaman, Mahmudul Hasan, Mohammad Nurul Huda · International Journal of Computer Applications · 2014
environmental sources of noise and distortion can degrade the quality of the speech signal in a communication system. This study explores the effects of these intrusive sounds on speech applications, introduces some techniques for reducing the influence of noise and enhances the acceptability and intelligibility of the speech signal. In our research, a noise reduction system incorporates a single microphone method in time domain to improve SNRs (signal to noise ratios) of noise contaminated speech. Previously, noise reduction techniques estimate noise from the valley of the spectrum based on the harmonic properties of noisy speech, called minimum value sequences (MVS). Since the valleys of spectrum are inadequate to estimate noise reliably, we propose the estimated degree of noise (EDON) (1), (2) to adjust the amplitudes of the MVS. The salient features of the proposed method are a single-channel adaptive filter to reduce computational time and cost for optimal noise reduction, and to estimate noise continuously on a frame-by- frame basis without the aid of voice activity detector (VAD) (2). For optimal noise reduction with a fewer number of iterations, an equation is derived from set values of SNRs and EDONs. To derive the proposed iteration number equation, we use the third degree parabola equation and least squares solution for the coefficients of EDON.