paper gives a novel approach of automatic speaker recognition technology, with an emphasis on text-dependent speaker recognition. Speaker recognition has been studied actively for several decades. In fact, Speaker recognition system may be viewed as working in four stages, namely, analysis, feature extraction, modeling and testing. After some preprocessing modules, we apply MFCC, as one of the most important feature extraction methods in this field of works, to speech signals independently in order to extract feature vectors. Afterwards, obtained vectors are used by training system to find codewords for ten users in our Persian database by LBG VQ. Finally, we use DTW technique for recognizing a speaker among all. Our experience strongly indicates that the identification rate over 96% can be achieved by the proposed algorithm. General Terms Speeches analyze,Speaker recognition.