- ai_shellv1
- thchs30
- aidatatang
- st-cmds
- primewords
- acoustic model is designed by dfcnn and ctc
- language model is designed by transformer encoder
python3 train.py
python3 test.py
- model is designed by transformer from filter bank to chinese
python3 transformer_train.py
python3 transformer_test.py