NVIDIA/Maya-ACE v2.0.0

v2.0.0

Repository: NVIDIA/Maya-ACE

Tag: v2.0.0

Published: 2025-09-24T16:36:30Z

Prerelease: no

Release notes:

Audio2Face-3D Local Inference Workflow

• Introduce the Audio2Face-3D Local Inference workflow via the Audio2Face-3D SDK with the latest generative AI models.
• Add support for all available Audio2Face-3D and Audio2Emotion models, including Audio2Face-3D v3.0 (diffusion-based, multi-identity).
• Add A2FAnimationPlayer node that provides interactive feedback and control over the generative models.
• Add geometry and transform outputs alongside blendshape weights, enabling the full capabilities of the models.
• Add sample Maya scenes and assets for the new workflow.
• Add TRT Model Manager to view and manage TensorRT models.
• Add documentation for the new workflow and instructions to download TensorRT models.
• Enable Import/Export A2F Parameters between AceAnimationPlayer and A2FAnimationPlayer.

Other New Features

• Add production example using MetaHuman and batch script.
• Support connecting an ACE/A2F Animation Player to multiple blendshape nodes.
• Add support for YAML configuration files.

Improvements and Bug Fixes

• Move and anchor the Request New Animation button to the top of the UI.
• Refine attribute names and the Attribute Editor UI of the ACE Animation Player.
• Move the Config Import/Export dialog to the center of the window, not the screen.
• Remove MEL script usage in the RequestSendAudio command.
• Unlock attributes before applying preset attributes.
• Upgrade OpenSSL to 3.0.16.

Deprecated

• Remove Audio2Face-3D Authoring service-related features and UI
• Remove Client Type option from Network Settings
• Remove interactive authoring mode in the ACE Animation Player
• Remove deprecated NVCF function IDs from presets
• Drop support for Maya 2023 and earlier