NVIDIA unveils ACE AI microservices for digital human creation

NVIDIA has announced the general availability of its ACE generative AI microservices, which aim to accelerate the next wave of digital humans. These technologies promise to facilitate the creation, animation, and operation of lifelike digital humans across various sectors, including customer service, gaming, and healthcare.

Companies such as Dell Technologies, ServiceNow, Aww Inc., Inventec, and Perfect World Games are among the first adopters of ACE technologies. The suite of digital human generative AI technologies now available includes NVIDIA Riva for automatic speech recognition (ASR), text-to-speech (TTS) conversion, and neural machine translation (NMT); NVIDIA Nemotron for language understanding and contextual response generation; NVIDIA Audio2Face for realistic facial animation based on audio tracks; and NVIDIA Omniverse RTX for real-time, path-traced realistic skin and hair animations.

Newly announced technologies include NVIDIA Audio2Gesture, which generates body gestures based on audio tracks, and the Nemotron-3 4.5B, a new small language model (SLM) designed for low-latency, on-device RTX AI PC inference. Jensen Huang, founder and CEO of NVIDIA, remarked that digital humans have the potential to revolutionise multiple industries. He also highlighted the breakthroughs brought by multi-modal large language models and neural graphics, which are set to pave the way for more intuitive human-computer interactions.

Previously, NVIDIA offered ACE as NVIDIA NIM microservices for developers to operate in data centres. However, the company is now expanding its reach with ACE PC NIM microservices set to be deployed across the 100 million RTX AI PCs and laptops installed worldwide. The Nemotron-3 4.5B is currently in early access, and NVIDIA Audio2Face and Riva ASR on-device models will soon follow. To simplify deployment, NVIDIA has introduced the NVIDIA AI Inference Manager software development kit, which preconfigures PCs with necessary AI models, engines, and dependencies, orchestrating AI inference across both PCs and the cloud seamlessly.

At the Computex trade show, attendees had the opportunity to experience ACE technologies first-hand with an updated version of the Covert Protocol tech demo. Developed in collaboration with Inworld AI, the demo used Audio2Face and Riva ASR technology running on GeForce RTX PCs, allowing players to interact with and influence digital-human non-playable characters through conversational language.

The ACE ecosystem is expanding rapidly, with an increasing number of developers from companies including Aww Inc., Dell Technologies, Gumption, Hippocratic AI, Inventec, OurPalm, Perfect World Games, Reallusion, ServiceNow, Soulbotix, SoulShell, and UneeQ building various applications. Aww Inc., which launched its first virtual celebrity Imma in 2018, is planning to use ACE Audio2Face microservices for real-time animation to create a highly interactive communication experience.

Perfect World Games is incorporating ACE technologies into its new tech demo, Legends, allowing players to interact with realistic, multilingual AI NPCs in both English and Mandarin. Inventec is utilising NVIDIA Audio2Face NIM to enhance its AI healthcare agent within the VRSTATE platform, offering a more engaging virtual consultation experience.

ServiceNow recently demonstrated the potential of ACE technologies in a generative AI service agent demo for its Now Assist Gen AI Experience, aiming to enhance customer and employee interactions across various industries. Dell Technologies also showcased its Dell Generative AI Solution for Digital Assistants at Dell Technologies World, designed to enable businesses to engage customers through natural conversations.

The NVIDIA art teams used generative AI tools, including Synthesia and Hour One, to create a digital avatar of founder Jensen Huang. The avatar, featuring Huang’s unique voice and style, was generated using ElevenLabs' AI speech and voice technology. It is available in Mandarin Chinese and English. Additionally, NVIDIA collaborated with Voicemod to compose the theme song for Huang’s keynote.

NVIDIA ACE NIM microservices for server deployments, including Riva and Audio2Face, are now in production, and developers can receive enterprise-class support.