Speaker Verification System using the combination of feature parameters based on C#

摘要

This thesis improved the traditional Speaker Verification System (SVS) by using a combination of feature parameters and a new method to calculate the threshold of a SVS is also presented. The improved system reduces false rejection rate and false acceptance rate. The system can furthermore record the voice of impostors who try to imitate the real speakers, thus increasing its security. The whole system is based on C# language.