International Journal of Artificial Intelligence and Neural Networks
Author(s) : ALI MOHAMMED BEN RAMADAN
In a world where authentication and privacy are taking a lot of our daily efforts, it is becoming more important for us to prove our identity to different systems every day so that we can access required and useful services by speaker verification system. Speaker verification (Is the speaker who we think he or she is ?) .In addition , speaker verification can be closed-set (The speaker is always one of a closed set used for training.) or open-set (speaker from outside the training set may be examined .). Also, each variant may be implement as text-dependent (The speaker must utter one of a closed set of words.) or text-independent (The speaker may utter any type of speech). In this project we explore the ability of a multilayer perception (MLP) to perform text-dependent speaker verification. Our networks are trained on sets of acoustical parameters extracted form samples obtained from a closed set of speakers uttering a set of known words .Our primary feature extraction tools are Mel Frequency Cepstral Coefficient (MFCC). All software for this project was created using Matlab Version 6.5 , and neural network processing was carried out using the Matlab toolbox .