基于Matlab的语音识别系统设计

Design of Speech Recognition System Based on MatlabAbstract:Due to the communication technology developing continuously,the technology of speechrecognition has also made extraordinary accomplishments,and the performance remands of thetechnology of speech recognition are increasingly getting higher.Speech recognition technology is atechnology that usually recognizes what people are talking about.Because of its security,efficiency,lowprice,and easy-to-implement functions,,it can be combined with other speech processing technologiesto create more complex and practical applications The importance for the development of computersand social life has become increasingly prominent.The subject studied in this paper is the research onthe design of speech recognition system based on Matlab,specifically the research on the speechrecognition design of designated people.This article mainly introduces and uses the two algorithms of Gaussian mixture model (GMM)andMFCC (Mel frequency cepstrum coefficient)to realize the process of speech recognition.Use theGaussian function to approximate the probability of the signal to the maximum extent.First,preprocessthe voice signal;next,extract the characteristic parameters of the voice over MFCC;build up a templatelibrary to store a large deal of voice signals that have extracted the characteristic parameters;comparethe voice signal to be identified with the voice in the template library The signals are compared afterpassing through the same channel,and the comparison result is finally obtained.As it is tough to see theconsequences in the program,so added the MATLAB GUI interface.The user can watch the basicprocess of voice recognition by operating the buttons in the MATLAB GUI interface,and the finalrecognition result will also be clearly displayed on the interface.After many practical operations in thisdesign,the accuracy ratio of speech recognition can reach about Ninety percent.Keywords:MATLAB GUI,MFCC,GMM