Development and implementation of speaker verification algorithms

This paper compares several speaker verification algorithms. Speaker verification algorithms are employed in many new electronic systems. Given a discrete utterance, we use a combination of feature extraction and parametric speech analysis to obtain a parameter set representing the utterance. Many such utterances of the same word are characterized in this way, and the extracted parameter sets are compared using pattern matching techniques. The evaluation will be based on how consistently patterns match for trials of the same speaker and how they differ for trials of different speakers.