In this paper, various feature extraction techniques for text independent speaker identification such as Mel-frequency cepstral coefficients(MFCC), Modified Mel-frequency cepstral coefficients(MMFCC), Bark frequency cepstral coefficients(BFCC), Revised Perceptual liner prediction (RPLP) and linear predictive coefficient cepstrum (LPCC) are implemented and the comparison is done based on performance and computation time. For modeling speaker identity vector quantization (VQ) codebook have been used. The feature extraction technique with maximum identification accuracy and less false acceptance rate is identified by varying initial centroids. The algorithms were compared using TIMIT database of 100 speakers.