Abstract: This paper describes the development and implementation of an automatic speaker recognition system. The system uses text independent speaker verification with MFCC features and GMM based speaker modelling for authenticating the user. The developed system has been used by the 52 samples of students on a regular basis. The system tends to have a very high accuracy of 99% with only 7.6 seconds per sample response time for a 10 second voice. The system only needed 10-20 seconds sample to accurately predict the user and also verify if the sample is from known voice or unknown voice also.

Keywords: User Identification, Speaker Identification, GMM.