End-To-End DNN Based Speaker Recognition Inspired by I-Vector and PLDA

doi 10.1109/icassp.2018.8461958