luomingshuang/icefall_vsr_grid_visualnet_ctc

Pre-trained VisualNet-CTC models for the GRID visual dataset with icefall.

The model was trained on full GRID with the scripts in icefall. See (https://github.com/k2-fsa/icefall/tree/master/egs/grid/AVSR/visualnetctcasr) for more details of this model.

How to use

See (https://github.com/k2-fsa/icefall/blob/master/egs/grid/AVSR/visualnetctcasr/Pre-trained.md)

Training procedure

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

• —Install k2 and lhotse, k2 installation guide refers to https://k2.readthedocs.io/en/latest/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/grid/AVSR
bash ./prepare.sh

• —Training

export CUDA_VISIBLE_DEVICES="0"
python visualnet_ctc_asr/train.py --world-size 1

Evaluation results

The best decoding results (WER) on GRID TEST are listed below, we got this result by averaging models from epoch 16 to 25, the decoding method is 1best. ||TEST| |--|--| |WER|15.68%|