Note: This recipe is trained with the codes from this PR k2-fsa#375
The model was trained on full Aidatatang_200zh with the scripts in icefall based on the latest version k2.
The main repositories are list below, we will update the training and decoding scripts with the update of version. k2: https://github.com/k2-fsa/k2 icefall: https://github.com/k2-fsa/icefall lhotse: https://github.com/lhotse-speech/lhotse
- Install k2 and lhotse, k2 installation guide refers to https://k2-fsa.github.io/k2/installation/index.html, lhotse refers to https://lhotse.readthedocs.io/en/latest/getting-started.html#installation. I think the latest version would be ok. And please also install the requirements listed in icefall.
- Clone icefall(https://github.com/k2-fsa/icefall) and check to the commit showed above.
git clone https://github.com/k2-fsa/icefall
cd icefall
- Preparing data.
cd egs/aidatatang_200zh/ASR
bash ./prepare.sh
- Training
export CUDA_VISIBLE_DEVICES="0,1"
./pruned_transducer_stateless2/train.py \
--world-size 2 \
--num-epochs 30 \
--start-epoch 0 \
--exp-dir pruned_transducer_stateless2/exp \
--lang-dir data/lang_char \
--max-duration 250
The decoding results (WER%) on Aidatatang_200zh(dev and test) are listed below, we got this result by averaging models from epoch 11 to 29. The WERs are
| dev | test | comment | |
|---|---|---|---|
| greedy search | 5.53 | 6.59 | --epoch 29, --avg 19, --max-duration 100 |
| modified beam search (beam size 4) | 5.27 | 6.33 | --epoch 29, --avg 19, --max-duration 100 |
| fast beam search (set as default) | 5.30 | 6.34 | --epoch 29, --avg 19, --max-duration 1500 |