NVIDIA Launches AI Virtual Image Generation Platform

On November 9th, NVIDIA announced the launch of the NVIDIA Omniverse Avatar platform for creating AI avatars. Previously, NVIDIA’s meta-universe concept was widely circulated in the industry. The AI ​​avatar introduced this time is related to the avatar in the meta universe concept from the product point of view.

Omniverse Avatar is a technology platform for generating interactive AI avatars. Connected with technologies in speech AI, computer vision, natural language understanding, recommendation engines, and simulation technologies. The avatar created in the platform is an interactive character with ray-traced 3D graphics that can see, speak, understand language, etc.

Omniverse Avatar will open the door to creating artificial intelligence assistants that can be easily customized for any industry. These can help handle billions of daily customer service interactions. For example, restaurant orders, bank transactions, personal appointments and reservations, etc.

The Omniverse Avatar is a part of NVIDIA Omniverse, a virtual world simulation and collaboration platform for 3D workflows. It is currently in public beta and has more than 70,000 users. In the keynote speech of NVIDIA GTC, three examples of Omniverse Avatar were shared:

Project Tokkio for customer support

NVIDIA DRIVE Concierge for vehicle intelligence services

Project Maxine for video conferencing

In the Project Tokkio demonstration, a customer service avatar for real-time conversations and restaurant sales was shown with a replica avatar. In the DRIVE Concierge AI demonstration, the digital assistant on the central screen of the dashboard helps the driver choose the best driving mode to arrive at the destination on time. Project Maxine showed a video call in a noisy cafe, where the speaker’s voice kept the same voice and intonation as hers, and was clearly heard, transcribed and translated into other languages ​​without background noise. 

Omniverse Avatar uses elements from voice AI, computer vision, natural language understanding, recommendation engines, facial animation and graphics, and is supported by the following technologies:

Speech recognition is based on NVIDIA Riva, a software development kit that can recognize speech in multiple languages. Riva is also used to generate human-like voice responses from text. Its natural language understanding is based on the Megatron 530B large-scale language model, which can recognize, understand and generate human language. Megatron 530B is a pre-trained model that can complete sentences with little or no training, answer a large number of questions in subject areas, summarize long and complex stories, translate into other languages, and handle many untrained field.

The recommendation engine is provided by NVIDIA Merlin, a framework that allows companies to build deep learning recommendation systems that can process large amounts of data to make more informed recommendations.

Perception is enabled by NVIDIA Metropolis, a computer vision framework for video analysis.

Avatar animation is supported by NVIDIA Video2Face and Audio2Face, 2D and 3D artificial intelligence driven facial animation and rendering technology.

Finally, these technologies are combined into an application, and use the NVIDIA unified computing framework for real-time processing. These skills are packaged into scalable, customizable microservices that can be safely deployed, managed, and orchestrated in multiple locations through NVIDIA Fleet Command.

Posted by:CoinYuppie,Reprinted with attribution to:https://coinyuppie.com/nvidia-launches-ai-virtual-image-generation-platform/
Coinyuppie is an open information publishing platform, all information provided is not related to the views and positions of coinyuppie, and does not constitute any investment and financial advice. Users are expected to carefully screen and prevent risks.

Leave a Reply