e-ISSN:0976-5166
p-ISSN:2231-3850


INDIAN JOURNAL OF COMPUTER SCIENCE AND ENGINEERING

Call for Papers 2024

Feb 2024 - Volume 15, Issue 1
Deadline: 15 Jan 2024
Publication: 20 Feb 2024

Apr 2024 - Volume 15, Issue 2
Deadline: 15 Mar 2024
Publication: 20 Apr 2024

More

 

ABSTRACT

Title : FILTERING OUT NOISE IN SPEECH SIGNAL WITH FIR FILTER BASED ONSINGLE COSINE TERMGENERALISED ADJUSTABLE WINDOW
Authors : Mbachu C. B, Akaneme S. A., Muoghalu C. N.
Keywords : Adjustable Window, single-cosine term, high frequency noise, power spectral density.
Issue Date : Dec 2019-Jan 2020
Abstract :
Speech signal is widely and effectively used globally for communication purposes in all sectors of human eandeavour. Incidentally speech signals suffer from interference by noise components when transmission through a communication channel is involved. Such noise components include Additive White Gaussian Noise (AWGN), Random Noise, power line noise, high and low frequency noise components. Without removing these noise components the speech signal quality will be compromised, implying delivery of unreliable message. In this paper a FIR filter is designed with single cosine term generalised adjustable window with to filter out high frequency noise components from speech signals. The optimal parameters of the filter are sampling frequency of 44100Hz and filter order of 34. A real voice statement, “Education is the Key to the Development of any Nation” from the lead author of this paper is captured with the in-built microphone of a laptop system, recorded in windows media audio (.wma) format and stored in one of the files of the system as speech. The optimal sampling frequency of 44100Hz is applicable of the type of format of the speech signal. With “audioread” instruction the stored speech is loaded into a matlab edit window. A noise component of 4500Hz and above is generated with matlab and added to the speech to imitate real corrupt speech signal. When the corrupt signal is passed through the designed filter the noise component was drastically filtered out.
Page(s) : 139-149
ISSN : 0976-5166
Source : Vol. 10, No.6
PDF : Download
DOI : 10.21817/indjcse/2019/v10i6/191006100