Publication

Exploiting semi-supervised training through a dropout regularization in end-to-end speech recognition