Generate realistic facial animation from audio using NVIDIA Audio2Face. Works offline in Unity Editor, supports ARKit blendshapes and AnimationClip creation.Audio Face Animator Lite is a Unity plugin that automatically generates realistic facial animation from audio files using the NVIDIA Audio2Face AI model.The plugin analyzes speech and produces synchronized facial movements driven by ARKit-compatible blendshapes, making it easy to animate characters.Audio Face Animator Lite includes the core functionality of the full version for evaluating the workflow. It processes audio clips up to 5 seconds and runs only in the Unity Editor. Support for standalone builds is available in the full version.FeaturesGenerate facial animation directly from audio files.Powered by NVIDIA Audio2Face AI technology.Fully offline processing — no internet connection required.Accurate audio-to-animation synchronization.Uses ARKit blendshapes for facial animation.Create and save Unity AnimationClips.Runs exclusively inside the Unity Editor on Windows.Supports audio clips up to 5 seconds in length.RequirementsWindows 10 / 11, 64-bit.NVIDIA GPU, compute capability 7.5 or newer — GeForce RTX 20-series or newer, or the equivalent Quadro / RTX A-series / data-center card. AMD and Intel GPUs are not supported.NVIDIA driver 576 or newer. CUDA and TensorRT are bundled with the plugin; nothing has to be installed separately.6 GB or more free VRAM while generating.About 3 GB of disk space, and roughly the same again while the plugin copies its model files into StreamingAssets.Character model with ARKit blendshapes.First runThe plugin ships the AI model as ONNX and compiles a TensorRT engine matched to your GPU the first time you generate. This takes about a minute on a modern GPU and happens once — the engine is cached and reused from then on. In the Editor you can do it up front from Tools > AudioFaceAnimator > Setup TensorRT Engine, which shows your GPU and a progress bar.The same applies to a standalone build: the first generation call on each player's machine builds an engine for that GPU, once, and caches it. Show your own loading UI while it runs. This is why the model ships as ONNX — a prebuilt engine only loads on the GPU architecture it was compiled for.Audio Face Animator Lite provides a fast and efficient workflow for creating expressive facial performances from voice recordings without manual keyframing.───────────────────────────────MORE FROM DEVELOOPReady for the full version? Audio Face Animator:https://assetstore.unity.com/packages/slug/383624All our Unity tools and animation packs:https://dev-eloop.github.io/Support: dev.eloop@outlook.comSupported Platform: Windows (Editor only)GPU Requirement: NVIDIA, compute capability 7.5+ (RTX 20-series or newer), driver 576+Processing Mode: Local/offline onlyAI Model: NVIDIA Audio2FaceFacial Rig: ARKit blendshapesInput Format: WAV or Unity AudioClip — converted to 16 kHz mono automaticallyOutput: Real-time blendshape animation and Unity AnimationClipsAudio Face Animator uses Nvidia's Audio2Face model to generate facial animations from audio. GitHub Copilot was used to assist with code refactoring and testing during development. All generated suggestions were reviewed, validated, and modified as necessary by the developer before inclusion in the final package.
