Performance Analysis of Speaker Identification System With and Without Spoofing Attack of Voice Conversion

The aim is to make machine inevitably to identify a person or feasibly to substantiate a person's claimed identity through their speech. This paper deals with the Speaker Identification System which uses MFCC in training phase for extracting features. The system is being analyzed with and without applying Voice Conversion Attack using GMM which is a part of spoofing attack. In testing phase, the extracted features are given to the speaker design and the decisions are obtained by comparing the extracted parameters which are stored in database with the speaker's voice parameter. The parameter being analyzed is Speaker Identification Characterization(SIC).