5 datasets found

The LRW dataset is an English language lip reading dataset, containing 500 different words, each spoken by over 1,000 persons.

A dataset for real-time lip localization and tracking for lip reading

LRW-1000 is a challenging Mandarin lip-reading dataset due to its large variations in scale, resolution, background clutter, and speaker attributes.

Lip reading aims to recognize speech content from videos via visual analysis of speakers' lip movements. This is a challenging task due to the existence of homophemes – words...

Lip reading in the wild: A large-scale audio-visual dataset for lip reading and audio-visual speech recognition.