From: Wise teachers train better DNN acoustic models
Network
# of Params
Data
Hub5’00-SWB
RT03S-FSH
6 ×2048 (hard align., sMBR)
30.4 mil.
110 h transcribed (SWBD)
18.3 %
22.6 %
5 ×512 (soft target-trained)
3.4 mil.
1100 h untrans. (SWBD + FSH)
18.0%
21.4%