Exploring Pre-Training with Alignments for RNN Transducer Based End-to-End Speech Recognition.

Hu Hu, Rui Zhao, Jinyu Li, Liang Lu, Yifan Gong

Browse the full ICASSP paper archive.