Sequence Training of Encoder-Decoder Model Using Policy Gradient for End-to-End Speech Recognition.

Shigeki Karita, Atsunori Ogawa, Marc Delcroix, Tomohiro Nakatani

Browse the full ICASSP paper archive.