November 12, 2023: In a significant advancement in video conferencing and editing technology, NVIDIA has unveiled the latest iteration of its Maxine developer platform.
This update introduces groundbreaking GPU-accelerated AI services in the cloud, reshaping the way developers and businesses enhance video and audio streams in real time.
NVIDIA Maxine, now exclusively available on NVIDIA AI Enterprise, brings to the fore a range of advanced features, including augmented reality, audio effects, and video effects, that are set to redefine user experience in digital communication.
One of the standout features is the Live Portrait option, enabling users to animate a selected photo with just a standard webcam, syncing head movements and facial expressions to create a dynamic avatar.
This feature, boasting high-resolution model outputs, adds a personal touch to digital interactions.
Another innovative aspect is the Voice Font feature, still in early access, which allows users to generate a unique digital voice from a 30-second audio sample. This tool is particularly beneficial for individuals with speech impediments or those looking to refine their vocal identity.
Using AI and webcam feeds, Maxine’s Eye Contact feature directs users’ gaze towards the camera, enhancing conversation engagement.
This feature has been refined to include natural micro-eye movements and an adjustable “look away” function for more realistic interactions.
The latest Maxine release also includes:
- 3D Body Pose for upper body pose estimation, adding depth to virtual interactions.
- Support for NVIDIA L4 Tensor Core and L40 GPUs, boosting performance by up to 1.75x.
- NVIDIA Triton Inference Server support for various features, improving throughput in both single- and multi-GPU workflows.
- Enhanced audio super resolution and Eye Contact performance.
NVIDIA’s partners, such as Quicklink and CoPilot AI, are already reaping the benefits of Maxine in professional settings. Quicklink has integrated Maxine into its Cre8 video production tool, enabling professional-grade virtual and hybrid events.
CoPilot AI leverages Maxine for content creation, empowering users to easily produce engaging videos.
Maxine’s flexible deployment options ensure high-quality communication from nearly any location, whether in the cloud, on-premises, or at the edge. The latest production release of Maxine, part of NVIDIA AI Enterprise 4.1, provides users with a suite of ready-to-use features, including the Triton Inference Server and enterprise support.
For those interested in the limited early access program of NVIDIA Maxine, applications are open online. NVIDIA also invites participants to contribute feedback for future enhancements through the NVIDIA Maxine and NVIDIA Broadcast App survey.
