Research promises to take video trickery a step further

Researchers in the US are working on using artificial intelligence to advance video lip-synching technology

Researchers in the US are working on using artificial intelligence to advance video lip-synching technology. The Massachusetts Institute of Technology (MIT) programme learns how individuals speak before putting words into their mouths.

The system is at an early stage of development but promises to be sophisticated enough to offer a range of applications.

Viewers in early trials were able to tell the real images from the animated ones only about 50 per cent of the time.

Aside from its obvious appeal to the entertainment industry, it could be used for language training or speech therapy.

The computer first studies a short video clip to learn how the person shapes their mouth. The process is painstaking and can take several days.

It then digitally morphs to shape of the subject's mouth around any audio sequence the creator wants them to speak.

The MIT team is aware of concerns that the work might be used for unethical reasons.

But its members say there's no more danger of misuse than with any other innovation.

They argue video watermarks could prevent copying or alteration of tapes in the same way banknotes are marked to prevent forgery. PA

  • Join The Irish Times on WhatsApp and stay up to date

  • Sign up for push alerts to get the best breaking news, analysis and comment delivered directly to your phone

  • Listen to In The News podcast daily for a deep dive on the stories that matter