Speaker and Gender Identification using Multilingual Speech

As the demand for multilingual speaker recognizers increases, the development of systems which combine automatic speaker and gender identification, models becomes increasingly important. In this work a speaker and gender identification system is developed using multilingual speech signal as input. MFCCs and delta-MFCCs, LPC, LPCC , Formants ,ZCR are used to build modal for classification and to reduce size of feature vector k-means clustering used. Radial basis function network and multi-layer perceptron are used for classification and their results are compared. Here resilient back propagation algorithm used to train MLP. Two separate modules are used for gender and speaker identification in each experiment. In this experiment accuracy of gender identification is 99% and speaker recognition is 91% using back propagation algorithm and 98% and 92% for gender and speaker identification using radial basis