A speaker recognition system using by cross correlation

In this paper, we studied a speaker independent isolated speaker recognition system for Turkish language by using cross correlation technique. The power spectrum's of each keyword speech for different speaker's determined using the linear predictive coding in order to constitute a feature vectors database that consists these spectrums. A new test input power spectrum's then compared with these feature vectors database by using cross correlation technique and obtained the cross correlation value's of each comparison. The results of matching scores evaluated in different SNR levels and different number of persons.