In this book, an improved strategy to the improvement of an automated text dependent speaker identification system as a biometrically-based technology has been studied and it is concerned with the close set text dependent speaker identification process using genetically optimized Hidden Markov Model with cepstral based features. At first, speech is taken by using microphone. Then some speech pre-processing techniques such as start and end points detection, silence part removal, pre-emphasis filtering, speech segmentation, windowing etc techniques have been applied. After pre-processing, features are extracted by different techniques to optimize the performance of the identification. RCC, MFCC, ΔMFCC, ΔΔMFCC, LPC and LPCC have been used to extract the features. To design the codebook, Genetic Algorithm has been used. Finally, HMM is used in the learning and identification phases. To remove the background noise, Wiener filter has been used. To measure the performance, a standard speech corpus NOIZEUS has been used. The experimental result shows the superiority of this proposed GA-HMM based close-set real time speaker identification system.