Nvidia has introduced Maxine, an AI-based video streaming platform for better quality in video calls. It is a tool for developers who can apply it to video calling services. It uses Nvidia’s cloud-based AI and GPU processing to improve call quality with added features.
The company says that Maxine can help developers reduce video bandwidth usage, live translate, and support audio-video effects. Since it is a cloud-based service, users can get the same features on their smartphones, laptops, and other devices.
What does Nvidia Maxine Do?
As mentioned, Maxine is a cloud-based video streaming platform to enhance video quality during calls and meetings. To explain further, it uses the computational capabilities of AI to reduce noise and add new features to a video call. Starting with what Nvidia calls ‘super-resolution’, it uses AI-backed artifact reduction to convert low-quality videos into high resolution in real-time.
It is using AI-based video compression, which reduces the bandwidth use to one-tenth of what is needed for the current H.264 video compression standard.
Maxine improves video resolution by transferring your face’s ‘keypoints’ over the internet instead of an entire screen of pixels. It means that the AI will sort the information it needs, and reconstruct it into a better-quality video at the receiver’s end.
Nvidia also posted a video showing off Denoise, and face re-animation features on Maxine. Denoise is an AI-backed noise cancellation feature aimed at amplifying your voice and reducing the background noises during a video call. Face re-animation automatically aligns your face, so it looks like you’re looking directly into the camera.
The re-animation also uses AI-recognized key points of a person’s face and re-animates it on the receiver’s end using generative adversarial networks (GANs). Re-animation can be used to stimulate eye-contact on a video call.
The platform also allows users to converse with Nvidia Jarvis and other conversational AI services. Much like Tony Stark’s AI with the same name, Jarvis is a conversation artificial intelligence that can do real-time translations and suggest captions during video calls.
Who is it for?
With the Nvidia Maxine, the company is counting heavily on its AI muscle. Maxine isn’t available for the end-consumers. It is a toolkit for developers to improve video calls on third-party software. The company has called on the developers to join the early access program.