Fetching the paper…

DFSMN-SAN with Persistent Memory Model for Automatic Speech Recognition · Around