TY  - RPRT
TI  - Tokenwise Contrastive Pretraining for Finer Speech-to-BERT Alignment in End-to-End Speech-to-Intent Systems
AU  - Vishal Sunder
AU  - Eric Fosler-Lussier
AU  - Samuel Thomas
AU  - Hong-Kwang J. Kuo
AU  - Brian Kingsbury
PY  - 2022
UR  - https://arxiv.org/abs/2204.05188
ID  - 2204.05188
ER  -