On the compression of recurrent neural networks with an application to LVCSR acoustic modeling for embedded speech recognition.

Rohit Prabhavalkar, Ouais Alsharif, Antoine Bruguier, Ian McGraw

Browse the full ICASSP paper archive.