Real-time conversational AI is a complex task, because it needs to understand voice, text, and language and reply, and this process needs to be completed accurately in a very short time.
On November 9, NVIDIA announced the launch of NVIDIA Omniverse Avatar , a technology platform for generating interactive AI avatars.
In the keynote speech of NVIDIA GTC , NVIDIA founder and CEO Huang Renxun shared various examples of Omniverse Avatar: Project Tokkio for customer support, NVIDIA DRIVE Concierge for online vehicle intelligence services, and Project for video conferencing Maxine .
In the first Project Tokkio demonstration, Huang Renxun showed the AI robot version of himself, the Toy Jensen Omniverse Avatar, which has Huang Renxun’s voice and the image of a toy, and discusses in-depth topics such as biology and climate science with experts. .
In the second Project Tokkio demonstration, Nvidia showed a customer service avatar in a restaurant kiosk . It was able to see two customers, talk to them in real time, and understand that they needed to order vegetarian burgers, French fries and drinks.
In the field of vehicle intelligent services, a digital assistant that looks similar to restaurant waiters appears on the central screen of the dashboard. Its responsibility is to help the driver choose the best driving mode to arrive at the destination on time, and set the car's mileage to be low according to the driver's requirements. A reminder at 100 miles.
In addition, Nvidia demonstrated Project Maxine for video conferencing. When attending a video conference in a noisy cafe, a lady can remove background noise simply and quickly, and what she says will be transcribed and translated into multiple languages in real time. The Omniverse Avatar is responsible for generating a virtual lip and intonation. Image.
According to Nvidia, "the dawn of intelligent virtual assistants has arrived." Collaborative robots and virtual assistants can be easily customized for almost any industry, helping to handle billions of daily customer service interactions-restaurant orders, bank transactions, personal appointments, etc. , Which brings more business opportunities.
In order to implement all the above examples, Omniverse Avatar uses the following technologies:
(1) Speech recognition is based on NVIDIA Riva, which can recognize speech in multiple languages and generate a human-like voice response through the "text-to-speech" function;
(2) Natural language understanding is based on the Megatron 530B large-scale language model, which can answer questions in a large number of fields and summarize long and complex stories;
(3) The recommendation engine is powered by NVIDIA Merlin Provided, it allows companies to build deep learning recommendation systems that handle large amounts of data;
(4) Perception capabilities are provided by NVIDIA Metropolis, which is a computer vision framework for video analysis;
(5) The image animation is supported by facial animation and rendering technology driven by 2D and 3D artificial intelligence.
It can be said that Nvidia's AI role in all walks of life is also to show off its skills in a high-profile and elegant manner.
These technologies are combined into an application program, and use the NVIDIA unified computing framework for real-time processing; at the same time, they are packaged into scalable and customizable microservices, which can be accessed through NVIDIA Fleet Command Deploy, manage, and orchestrate securely.
Omniverse Avatar is NVIDIA Omniverse a part of. NVIDIA Omniverse It is a virtual world simulation and collaboration platform for 3D workflows. It also includes Omniverse Replicator, a synthetic data generation engine used to train deep neural networks, helping developers create the massive amounts of data needed to train AI.
NVIDIA Omniverse It is currently in the public beta and has more than 70,000 users. Huang Renxun said:
With Omniverse, we now have the technology to create a new 3D world or model our physical world. You will see an unchanging theme-how Omniverse is used to simulate warehouses, factories, physical and biological systems, 5G frontiers, robots, self-driving cars, and digital twins of avatars.
#Welcome to follow Aifaner's official WeChat account: Aifaner (WeChat ID: ifanr), more exciting content will be provided to you as soon as possible.