audio-visual models

Researchers have created a new audio-visual model that harnesses the power of deep neural networks that can isolate and enhance speech under natural conditions.