A MATLAB based improved Speech Enhancement and Recognition using Spectral Subraction Method,MFCC,GMM

Speech recognition is the process of automatically recognizing a particular word from a particular speaker based on the information contained in each voice wave. This document reviews inventions and technological advances in the field of speech recognition and also focuses on the various steps required to identify speaker using MATLAB programming. In this document, we first apply Spectral Subtraction Method to perform speech enhancement and noise removal. This background noise is successfully removed by the Wiener Filter application. In addition, the technique adopted here develops code using MATLAB programming. In this article, nine audio samples were recorded through a microphone and the system was trained according to the recoded audio samples. The MFCC function of the voice sample was calculated and the words were distinguished according to according to the energy associated with each sampled word and recognized using GMM.