VOICE SIGNAL SEPARATION FROM TRAIN NOISE CORRUPTED SIGNAL USING NOTCH AND ELLIPTIC FILTER

Sources of noise are sound generated by the engine, heat noise, wind noise, etc. In telecommunication the noise that effects much more to the speech signal is additive white Gaussian noise. So, analysis of speech signal is an important factor for the telecommunication industries. And reconstruction of the original speech signal from the noisy speech signal passing through the channel is a very difficult task. There are many digital and analog filters which are used for the removal of the noise signal from speech signal. Each filters have their advantages and disadvantages. Depending on our need we choose a suitable filter for de-noising the speech signal. In this paper, we focus on the noise in the speech signal due to the high frequency horn of the diesel or electric locomotives and many other external noises. First, we perform the analysis of the spectrum by the method known as Discrete Fourier Transform (DFT) using Fast Fourier Transform (FFT) algorithm and we performed an audio de-noising technique based on few types of analog filters like – elliptical filter, notch filter (of different type). We deduce an algorithm for the given filters and simulate it through the MATLAB platform. The results of MATLAB simulation are then analyzed to give better performance for de-noising speech signal.