NVIDIA ACE Brings Lifelike Digital Humans to Developers
Joerg Hiller Jun 06, 2024 14:55
NVIDIA ACE, using generative AI, is now available for developers.
NVIDIA has announced that its suite of technologies, collectively known as NVIDIA ACE, is now generally available for developers. This suite is designed to bring digital humans to life using generative AI, promising advancements in areas such as gaming, customer service, and healthcare.
ACE Now Available for Production Deployment
According to the NVIDIA Technical Blog, ACE is packaged as NVIDIA NIMs (Neural Inference Microservices). These microservices enable high-quality natural language understanding, speech synthesis, and facial animation. Leading companies, including Aww Inc, Dell Technologies, Gumption, Hippocratic AI, Inventec, OurPalm, Perfect World Games, Reallusion, ServiceNow, SoulBotix, SoulShell, and Uneeq, are integrating ACE into their platforms.
NVIDIA has also introduced ACE PC NIM microservices for deployment across the installed base of 100 million RTX AI PCs and laptops, available through early access.
Components and Updates
NVIDIA ACE 24.06 introduces general availability for various components within the digital human technologies suite, including NVIDIA Riva, NVIDIA Audio2Face, and NVIDIA Omniverse RTX. These are available through NVIDIA AI Enterprise.
Key microservices available in the NVIDIA NGC Catalog and the NVIDIA ACE GitHub repository include:
- Riva ASR 2.15.1: Adds a new English model with improved quality and accuracy.
- Riva TTS 2.15.1: Enhances representation for multiple languages and includes the beta release of P-Flow for voice adaptation.
- Riva NMT 2.15.1: Introduces a new 1.5B any-to-any translation model.
- Audio2Face 1.011: Adds blendshape customization options and improved lip sync for Metahuman characters.
- Omniverse Renderer Microservice 1.0.0: Adds new animation data protocol and endpoints.
- Animation Graph Microservice 1.0.0: Supports avatar position and facial expression animations.
- ACE Agent 4.0.0: Adds speech support for custom RAGs and prebuilt support for RAG workflows.
Early-access microservices include:
- Nemotron-3 4.5B SLM 0.1.0: Designed for on-device inference with minimal VRAM usage.
- Speech Live Portrait 0.1.0: Animates a person's portrait photo using audio.
- VoiceFont 1.1.1: Reduces latency for real-time use cases and supports concurrent batches across GPUs.
Developer Tools and Workflows
To facilitate the integration and deployment of ACE technologies, NVIDIA has released new workflows and developer tools available on the NVIDIA ACE GitHub. The Kairos Gaming reference workflow features an Audio2Face plugin for Unreal Engine 5, and the NVIDIA Tokkio customer service reference workflow includes various tools and samples for digital human configurations.
Additional developer tools include:
- Unified Cloud Services Tools 2.5: Streamlines NVIDIA Cloud Functions application deployment.
- Avatar Configurator 1.0.0: Adds a new base avatar, hairstyle, and clothing options.
ACE NIM Microservices for RTX AI PCs
Besides data center deployment, NVIDIA is bringing ACE NIM microservices to the installed base of 100 million RTX AI PCs and laptops. The first small language model, NVIDIA Nemotron-3 4.5B, is designed for on-device inference with comparable accuracy to large language models running in the cloud. Early access for Nemotron-3 4.5B SLM is now available, with Audio2Face and NVIDIA Riva ASR on-device models to follow soon.
The Covert Protocol tech demo, developed in collaboration with Inworld AI, showcases the capabilities of Audio2Face and Riva ASR running locally on GeForce RTX PCs.
Getting Started
Developers can begin working with NVIDIA ACE by evaluating the latest ACE NIMs directly from their browser or through API endpoints running on a fully accelerated stack. NVIDIA offers tools and workflows to accelerate integrations and early access to microservices to help developers see how ACE can transform their pipeline in the future.
For enterprises seeking an end-to-end digital human solution or custom development on ACE, NVIDIA Partner Network includes service delivery partners such as Convai, Inworld AI, Data Monsters, Quantiphi, Soulshell, Top Health Tech, and UneeQ.
For questions or feedback about digital human technologies, developers can visit the Digital Human forum.
Image source: Shutterstock