0:00:15a lot easier or a improving the robustness of speaker verification systems against maybe speech
0:00:23then i'll
0:00:24tested with a
0:00:26the original voice of the impose that the speaker state-of-the-art speaker verification systems perform better
0:00:32but that the impersonator is mimicking the target speaker the performance deteriorates significantly
0:00:42be addressed this issue you know in our work
0:00:46we use a proximal support vector machine based backend classifier its use i-vectors that input
0:00:52for improving the robustness against a mimicked speech
0:00:58then be devised as tight as it do maybe each of the training examples for
0:01:04this em to further improve the robustness of the system against mimicry it's is the
0:01:11or stuff that was not but you
0:01:14thank you