av_hubert
av_hubert copied to clipboard
Update hubert_dataset.py
change string based offset caculation to 'bytes' based offset caculation due to NON-ASCII language character label.(I'm working on Korean AVSR)