Speaker recognition with intelligent methods

In this study, text depended speaker recognizer system; based on Neural Network (NN) and Adaptive Network Fuzzy Inference System (ANFIS) are designed by using YOHO database. In feature extracting stage of speaker recognizer, MFCC and Wavelet Transform (WT) entropy values were used. The obtained feature vectors are used as inputs to the NN and ANFIS models. In test stage, recognition of the speakers is performed from the given speech samples. At the same time, MFCC feature vectors and wavelet transforms entropy values are used for examining the NN and ANFIS model parameters effects on the systems performance and finally the results are compared.