NVIDIA has released its latest update to the Maxine developer platform, which it believes will ‘redefine’ video conferencing.

Through this release, the AI hardware and software supplier is providing GPU-accelerated AI cloud services that can improve audio, video, and augmented reality in real-time.

NVIDIA Maxine is exclusively available on NVIDIA Enterprise and includes Live Portrait, Voice Font, Maxine Eye Contact, 3D Body Pose, audio resolution performance increases, and more.

The developer-centred videoconferencing solution, NVIDIA Maxine, was first launched in 2020 to solve video calling issues using artificial intelligence.

Live Portrait

NVIDIA’s Live Portrait feature allows users to animate photos of themselves using a standard webcam.

Live Portrait will sync the person’s head movements and facial expressions to this photo. Users can also create 2D stylised caricatures of themselves.

The Live Portrait feature is now available as a production feature in Maxine with high-resolution model outputs of up to 1024 x 1024.

Voice Font

Voice Font, available for evaluation and testing in the early access release of Maxine, lets users generate a unique voice to represent themselves. You can think of it like a digital avatar for voice.

The feature converts audio samples into a digital voice with only 30 seconds of audio reference material.

This can be particularly useful for people with speech impediments and those who would like to fine-tune the sound of their voice.

Maxine Eye Contact

Earlier in the year at CES 2023, NVIDIA introduced a new Eye Contact feature, which uses AI technology to make it appear as though the speaker is looking into the camera.

Now, NVIDIA has released a new version of Maxine Eye Contact that maintains naturally occurring micro-eye movement, with the added ability to set recurring ‘looking away’ intervals to create a more natural video conferencing experience.

The new Eye Contact feature also brings quality improvements, such as increased gaze stability, greater occlusion handling, and lower latency when used with NVIDIA Ada Generation GPUs.

According to NVIDIA, studies have shown that regular eye contact encourages personal connections, understanding, and engagement.

Further Capabilities

NVIDIA Maxine has been updated to include 3D Body Pose, which estimates body poses when only the upper body can be seen by the camera.

There are audio resolution performance improvements, with throughput gains of as much as 109 percent.

Support is available for NVIDIA L4 Tensor Core and L40 GPUs, which increases performance by 1.75x compared to GPUs from previous families.

There is also NVIDIA Triton Inference Server support for AI Green Screen, Eye Contact, Landmark Detection, and Face Detection, allowing higher throughput for single and multiple GPU workflows.

Copilot AI, an AI and sales enablement startup and NVIDIA customer, provided its feedback on Maxine. Jackson Chao, cofounder of CoPilot AI Video, said: “Using NVIDIA Maxine microservices, we empower users to record on the whim and stand out in the crowd.

“Maxine’s Eye Contact feature allows users to record a script without compromising the connection with viewers.

“Even novice content creators are able to adopt video outreach with confidence and humanize the way they engage with their contacts.”

Last month, NVIDIA and Cisco partnered on an AI-powered room kit, Room Kit EQX.

Unveiled at WebexOne in Anaheim, California, the kit will be underpinned by NVIDIA’s Jetson Edge AI platform.

Cisco has promised easy deployment, enhanced video and sound, and improved meeting operability with Microsoft Teams.

 

 



from UC Today https://ift.tt/hWRfKJi