Summary
This lets you generate talking head videos from a portrait photo and either text or audio. The recommended path is P-Video-Avatar, which has built-in TTS and runs 18x faster and 6x cheaper than alternatives like OmniHuman or Fabric. You can use it for product demos, explainer videos, UGC-style ads, or dubbing content into other languages. The workflow is straightforward: pass in an image URL and either a voice script (it handles TTS) or an audio file, and you get back a video with synced lip movements. It supports 30 voices across 10 languages and outputs up to 1080p. Good for anyone who needs to generate presenter-style videos at scale without filming real people.
Install
npx skills add https://github.com/inference-sh/skills --skill ai-avatar-video
