Handling i-vectors from different recording conditions using multi-channel simplified PLDA in speaker recognition | IEEE Conference Publication | IEEE Xplore