REDUCTION OF NOISE IN SPEECH SIGNAL PROCESSING AND RECONSTRUCTION OF SIGNAL

Noise is a signal where information is converted into voltage variation. The cause of noise is another signal of voltage variation which has occurred from a source which is not authorised known as unwanted or undesired signal. It may enter our communication system in various stages through a microphone from external sources such as wind, human disturbances, etc. This paper dilineates the LMS adaptive method used for audio quality amplification by minimising the noise in speech signal processing. The least mean square and leaky least mean square were the two basic adaptive algorithms has been executed to turn down the noise within the range of speech signal. This combined system has more efficiency and can be used in various places such as auditorium, etc for various purposes where the noise is the major problem. The software can be used easily to detect and minimize the noise using applications. Widrow's adaptive noise cancellation technique with some improvements in the matlab code and the signal to noise ratio improvements were presented in this paper.