issues
search
hirofumi0810
/
neural_sp
End-to-end ASR/LM implementation with PyTorch
Apache License 2.0
589
stars
140
forks
source link
Add safeguard for state reset during streaming inference
#303
Closed
hirofumi0810
closed
3 years ago
hirofumi0810
commented
3 years ago
Add safeguard
Add
dualhyp
option
Add backoff for LSTM encoders
Fix
key_tail
when block size is smaller than chunk size in MoChA
dualhyp
optionkey_tail
when block size is smaller than chunk size in MoChA