WFST-based structural classification integrating dnn acoustic features and RNN language features for speech recognition.

Quoc Truong Do, Satoshi Nakamura, Marc Delcroix, Takaaki Hori

Browse the full ICASSP paper archive.