October 16, 2021

Nvidia’s new technology will improve the quality of video calls using AI

Nvidia has announced a new video conferencing platform for developers called Nvidia Maxine, which it claims could improve the quality of communication.

According to The Verge, Nvidia Maxine is not a consumer platform, but a set of tools for third-party firms to improve their own software.

Maxine will handle calls in the cloud using Nvidia GPUs and enhance their call quality using artificial intelligence.

Using AI, the system can:

  • Rebuild the faces and views of the subscribers so that they always look directly into their camera;
  • Reduce video bandwidth requirements by transmitting only “face cues”;
  • Increase video resolutions higher than original;
  • Add artificial lighting to the face;
  • Create translation and transcription in real time;
  • Animate user avatars.

Of course, not all of these features are new. Video compression and real-time transcription are common enough, and Microsoft and Apple have implemented eye alignment in Surface Pro X and FaceTime to help people maintain eye contact when making video calls (although Nvidia’s tech looks better).

Nvidia hopes its edge in cloud computing and impressive research and development in artificial intelligence will help it outperform its competitors.

During a conference call, Nvidia’s general manager Richard Kerris described Maxine as “a really exciting and very timely solution” and singled out AI-powered video compression as a particularly useful feature.

“We’ve all had instances where bandwidth has been a constraint on our daily calls,” Kerris said. “If we apply AI to this problem, we can reconstruct the difference scenes at both ends of the call and only transmit what needs to be transmitted, thereby significantly reducing bandwidth requirements.”