Handling i-vectors from different recording conditions using multi-channel simplified PLDA in speaker recognition | Publicación