In fact, according to the scientists, the software can take a live video feed of a person talking and make them look and sound like somebody else could actually change that, the ‘New Scientist’ reported. In their research, the scientist recorded video of volunteers performing 30 different facial expressions such as frowning, smiling and looking surprised. For each expression, the positions of key facial features, such as the eyes, nose and corners of the lips, were manually labelled.

That annotated footage was used to “train” software to recognize the face of each individual featured in the set. Once trained on a person in this way, it can closely track every move of their face in video footage.

Those movements can then be transferred onto the face of another “known” person by calculating how the recipient’s features need to change to take on each new expression.