End-to-End Encoder Decoder models for ASR
End-to-end speech recognition using encoder-decoder models with auxiliary tasks on lower layers. The code is still undergoing refinement and the code to prepare data files etc would take some more time.
Publication(s):
Multitask Learning with Low-Level Auxiliary Tasks for Encoder-Decoder Based Speech Recognition
Interspeech, 2017
Shubham Toshniwal, Hao Tang, Liang Lu, Karen Livescu