High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Impulsive Noise Removal from Speech Signals using Rank Order Mean


Vijay Laxmi Shukla , Suyash Institute of Information Technology Gorakhpur,; Mr. Atul Sinha , Suyash Institute of Information Technology Gorakhpur,


Signal to Noise Ratio (SNR), Evaluation of speech Quality (PESQ), Frequency range of 20Hz to 20kHz, Bandwidth of only 4kHz etc


In this Paper represent the impulsive noise Removal speech using rank order mean. Speech is a very rudimentary way for persons to transport information to one additional with a bandwidth of only 4 kHz; speech can convey information with the feeling of a humanoid voice. The language indication has convinced properties: It is a one-dimensional signal, with time as its self-governing variable, it is chance in countryside, it is non-stationary, and i.e. the frequency spectrum is not continuous in time. Though humanoid existences have an audible frequency range of 20Hz to 20 kHz, the humanoid speech has important occurrence components only up to 4 kHz. In this work a method for removal of impulse noise from the speech signal using Rank Order Mean is proposed. The rank order differentiation is applied to input signal to estimate the time occurrence of impulsive noise. Then rank order mean is used for replacing the noisy samples to get the noise free signal. The above described technique shows improvement in terms of Signal to Noise Ratio (SNR) and Perceptual Evaluation of Speech Quality (PESQ) w. r. t to the existing techniques. Noise cancellation is the process of removing background sound from language sign. The squalor of speech due to presence of contextual noise and several other noises reason problems in various sign processing errands like speech credit, speaker recognition, and speaker verification etc. Numerous methods have been extensively used to remove noise from speech signal like linear and nonlinear filtering methods, adaptive noise annulment, total difference demising etc. This paper addresses the problem of reducing the impulsive sound in talking signal using compressive sensing method. The results are compared against three well known speech improvement means, spectral deduction, Total difference denoising and signal dependent rank order mean algorithm.

Other Details

Paper ID: IJSRDV4I20865
Published in: Volume : 4, Issue : 2
Publication Date: 01/05/2016
Page(s): 944-946

Article Preview

Download Article