JFA-based front ends for speaker recognition | IEEE Conference Publication | IEEE Xplore