Video recovery examples from LRS3-TED dataset

LRS3-TED contains examples from TED talks of various speakers. The following samples are drawn from V2S models trained on LRS3-TED.

Input video (silent)

Ref. video

ReVISE

DiVISe

Input video (silent)

Ref. video

ReVISE

DiVISe

Input video (silent)

Ref. video

ReVISE

DiVISe

Input video (silent)

Ref. video

ReVISE

DiVISe