fudan-generative-vision/hallo

Health v2 · maintenance only

Adoption not part of the score

8664 stars · 1120 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Hallo is a Python research library implementing hierarchical audio-driven visual synthesis for animating portrait images into talking-head videos. It uses diffusion-based models to generate lip-synced, expressive facial animation from a single reference image and an audio clip.

Use cases

• animate a portrait photo with audio to make a talking head video
• generate lip-synced video from a single image and speech audio
• create talking avatar videos from still photos
• make a digital human speak from an audio file
• research audio-driven face animation with diffusion models

When to choose

• you need to animate a single portrait image into a talking video from audio
• you want a research-grade open-source talking-head generation model
• you have GPU resources and want to run inference locally or via HuggingFace

When to avoid

• you need real-time lip sync for live streaming or video calls
• you lack a GPU and cannot run heavy diffusion inference
• you need full-body or multi-person video animation rather than portraits

Facets

library · maturity active

video-processing machine-learning deep-learning audio-processing image-processing artificial-intelligence computer-vision deep-learning python talking-head portrait-animation audio-driven diffusion-models lip-sync generative-ai video audio gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="fudan-generative-vision/hallo")