Text Dependent Speaker Verification System using Discriminative Weighting Method and Artificial Neural Networks

M.Z. Ibrahim, M. Khalid, and R. Yusof (Malaysia)


Modelling, speaker recognition, artificial intelligence, vector quantization and neural networks.


Speaker recognition is a process to recognize someone by their voice. The goal of speaker recognition is to extract, characterize and recognize the information about speaker identity. In this paper, we discuss both conventional and Artificial Neural Network (ANN) approach to speaker recognition system. The proposed system comprises three main modules, a feature extraction module to extract necessary features from speech waves, a Vector Quantization (VQ) module to generate the speaker codebook and an ANN module to classify the speakers based on their high discriminative power. The proposed intelligent learning system has been applied to a case study of text-dependent speaker recognition system and the performance is evaluated by applying two types of feature extraction techniques: Mel Frequency Cepstral Coefficients (MFCC) and Linear Predictive Cepstral Coefficients (LPCC). Experiment shows that the new proposed system provides significantly higher performance compare to conventional method.

Important Links:

Go Back