DIGITAL ENHANCEMENT OF THE QUALITY OF A NOISY RADIO NIGERIA NEWS BROADCAST

Abstract

This work presents the results of applying digital techniques to study and improve the quality of Federal Radio Corporation of Nigeria (FRCN) news broadcast signals. Until May 2008, the major FRCN live news broadcast was of very unsatisfactory quality and characterized by noisiness, poor intelligibility and distortion. While many solutions have been tried in the past, improvements were usually short-lived because of inadequate information about the technologies and the inevitable high rate of obsolescence of cutting edge technologies that tended to be adopted by anxious executives. This work is an attempt at employing proven "state of the a11"technologies within an indigenous implementation initiative. During the course of the study, a noisy broadcast signal was digitized and captured onto a computer and processed using MATLAB® Software (environment). Important characteristics including the spectral features of both signal and channel noise were determined. Following a survey, a spectral subtraction method which uses the properties of the noise to enhance quality was implemented. Subsequently, the statistical properties of the background and channel noise of the environment were determined. The results indicated that it was very close to synthetic and available white noise. Experiments to extract signals from signal and additive noise mixes verified that the spectral subtraction method could improve the signal to noise ratio by 15dB.The method was applied to the real-life broadcast and a significant improvement in quality was observed.