Link to the dataset: https://spandh.dcs.shef.ac.uk/gridcorpus/ update the file path according to where you store your dataset Link to the paper: https://arxiv.org/abs/1611.01599 If you want you can also try using LRW dataset. Link to the LRW dataset: https://www.robots.ox.ac.uk/~vgg/data/lip_reading/lrw1.html You have to mail them before using it.