![]() | Only 14 pages are availabe for public view |
Abstract Three proposed techniques for speaker identification under reverberation effect have been presented. First proposed used comb filter to represent the reverberation. Second proposed technique is cancelable speaker identification system using comb filter to distort the features of the speech before saving in database. Third proposed is cancelable speaker identification system that based on optimized version of the comb filter using zero cancelation. . The aim of this thesis is to study the effect of reverberations on the performance of speaker recognition systems, achieving safety for the databases, protecting the features of the speech from the hackers and increase the durability of the system by intentional distorting the features of the process speech using the comb filter and another optimized version of the comb filter. |