Now-a-days synthetic voice is frequently used to defraud a biometric access systems which is speaker recognition based. This paper presents synthetic speech detection in automatic speaker verification system (ASV) for the purpose of spoof detection. Feature extraction is done by canonical Mel Frequency Cepstral Coefficients (MFCC) algorithm and classification of natural and synthetic voice is done using Support Vector Machine (SVM). Several experiments are carried out, showing that nonlinear SVM performs better than linear SVM.