ADLSET

ADLSET contains a collection of audio-4D landmark pairs captured from one specific male subject. It contains about 80𝑚𝑖𝑛 news audio spoken in Mandarin. The news from the THUCNews dataset includes sports, real estate, entertainment topics, etc. For the landmarks, It contains 13 cheek landmarks and 12 outer lip landmarks. Users can use plot_lm.py file to plot the landmarks and get the animations.