De-noising the Speech Signal With FIR Filter Based on Matlab

Abstract

With the advent of the information age, signal transmission has been involving many kinds of fields. During the process of voice signal transmission, there would be inevitably doped with some noise signal, but which would not happen if the very designed filter has been used. In this paper, the window function method is used to select the fast Fourier transform interval, so as to design a FIR digital low-pass filter to achieve the doping in the speech signal in the random noise elimination, and analyze the time domain and frequency domain characteristics of voice signal before and after denoising, compare the quality difference of voice signal before and after filtering. This paper aims to verify the validity of the designed FIR digital low-pass filter's function by Matlab simulation. IntroductionVarious parts in modern communication systems are inseparable from the different forms of signal transmission. During the actual signal processing, signal generation, the completion of digital-to-analog conversion, and signal transmission process will be more or less affected by the impact of the external environment and then bringing the distortion. And most of the distortion is very serious, so that there would be great difficulties for signals to recover and then hide in the noise.This problem has a huge impact on the transmission quality of the communication system. Therefore, how to remove the noise off the signal is a key point to solve.This paper mainly discusses how to design a suitable FIR filter by Matlab software, so as to achieve better filtering effect. FIR filter The characteristics of FIR filterFIR filter is a finite impulse response filter, also known as non-recursive filter, is the most basic component in digital signal processing system. Its biggest feature is a very strict linear phase, and can guarantee any amplitude-frequency characteristics. At the same time its unit sampling response is finite, so which can be confirmed that the FIR filter is a stable system.Precisely because its unit impulse response is finite length, fast Fourier transform (FFT) algorithm can be used to filter signal, which can greatly improve the operation rate. FIR filter designing methodGenerally, the essence in designing process of FIR filter is that in accordance with the requirements of the frequency characteristics to achieve the frequency characteristics of indicators with the corresponding algorithm. Specifically, we use a certain approximation criterion to select each parameter h(n) in the filter transfer function H(z), that is to calculate the unit sampling response and the order N based on the optimal clamping criterion, so as to satisfy the required frequency characteristic's requirements. Common FIR digital filter designing methods are divided into the following three: window function method, frequency sampling method and Chebyshev approximation method.Among these three methods, the window function method is the most widely used, and the simplest. It can be calculated through the known window function and then use Ma...

Articulation Battery Test For Tamil Speech Disorder Signals

“…Acoustic speech signals may contaminate with noisy signals that can be removed by applying noise filtering techniques. Here, the popular filtering methods involve moving average filter, savitzky-golay filter, low pass filter, gaussian filter and butterworth filter are utilized to obtain the filtered speech signals [10][11][12][13][14]. The visual assessment of the filtered speech signals given by aforementioned filtering techniques are depicted in Figure7.…”