期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2015
卷号：3
期号：8
DOI：10.15680/IJIRCCE.2015. 0308050
出版社：S&S Publications
摘要：Voice disguise has shown an increasing tendency for criminal activities like kidnapping, threateningcalls. The most common way of disguising the voice adopted by the criminals is changing the pitch, so that speakerrecognition will be difficult. In this paper, we focus on improving the detection performance of disguised voice, whichis disguised by using software tools like Audacity, Cool Edit and Praat. First and foremost, we will be extracting melfrequencycepstral coefficient as acoustic feature. Then we will be using Probabilistic Neural Network (PNN) classifierto detect whether the voice is disguised or not. Here we will be showing the output of detection performance of SupportVector Machine (SVM) classifier for various disguising factors, which is the existing system and also the output ofPNN classifier for the disguising factor of -8 as of now.
