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

https://arxiv.org/abs/1704.01631