yerfor/GeneFace

Health v2 · maintenance only

Adoption not part of the score

2657 stars · 291 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

GeneFace is the official PyTorch implementation of an ICLR 2023 paper on generalized, high-fidelity audio-driven 3D talking face synthesis using NeRF-based rendering. It generates lip-synced talking head videos from arbitrary audio input, with pre-trained models and a successor project (GeneFace++) available.

Use cases

• generate a talking head video from an audio clip
• create lip-synced avatar videos driven by speech
• synthesize 3D talking faces with NeRF
• drive a person-specific avatar with out-of-domain audio
• research audio-driven facial animation
• build a digital human presenter from a reference video

When to choose

• you need research-grade audio-driven talking face synthesis with good lip sync
• you want a NeRF-based 3D talking head with pre-trained models to start quickly
• you are reproducing or extending the ICLR 2023 GeneFace paper

When to avoid

• you need real-time production performance or the best quality - use GeneFace++ or MimicTalk instead
• you lack a GPU or cannot handle a heavy PyTorch/3D reconstruction setup
• you need a simple 2D face-swap or lightweight avatar tool

Facets

library · maturity maintenance

deep-learning machine-learning video-processing audio-processing speech-recognition computer-vision deep-learning computer-vision artificial-intelligence python talking-face-generation nerf audio-driven pytorch face-synthesis lip-sync research-code iclr-2023 audio video linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="yerfor/GeneFace")