Fuzzy kernel vector quantization with entropy and sectional set for speaker recognition under limited data

In case of limited data, the system performance of speaker recognition decreased significantly. To resolve this problem, it designed fuzzy kernel entropy vector quantization with sectional set to train speakers´ models and make identification decision in high-dimensional feature space. Entropy function can make the algorithm have clear physical meaning and avoid the unsuitable choose of fuzzy weighted exponent. Sectional set method was used to modify the membership function, which can improve the convergence speed and recognition rate. Experimental results show that for about 5s of training and 1s of testing data, the performance of proposed method are 95.95%.