Performance Comparison of Automatic Speaker Recognition using Vector Quantization by LBG KFCG and KMCG

In this paper, three approaches for automatic Speaker Recognition based on Vectorquantization are proposed and their performances are compared. Vector Quantization(VQ) is used for feature extraction in both the training and testing phases. Threemethods for codebook generation have been used. In the 1st method, codebooks aregenerated from the speech samples by using the Linde-Buzo-Gray (LBG) algorithm. Inthe 2nd method, the codebooks are generated using the Kekre’s Fast CodebookGeneration (KFCG) algorithm and in the 3rd method, the codebooks are generatedusing the Kekre’s Median Codebook Generation (KMCG) algorithm. For speakeridentification, the codebook of the test sample is similarly generated and compared withthe codebooks of the reference samples stored in the database. The results obtainedfor the three methods have been compared. The results show that KFCG gives betterresults than LBG, while KMCG gives the best results.