Common vector approach and its combination with GMM for text-independent speaker recognition

摘要

In this paper, the common vector approach (CVA) is newly used for text-independent speaker recognition. The performance of CVA is compared with those of Fisher’s linear discriminant analysis (FLDA) and Gaussian mixture models (GMM). The recognition rates obtained for the TIMIT database indicate that CVA and GMM are superior to FLDA. However, while the recognition rates obtained from CVA and GMM are identical, CVA enjoys advantages in terms of processing power and memory requirement. In order to obtain better results than those achieved with GMM, a new method which is a combination of CVA and GMM is proposed in this paper.