OmniHuman 2.0 converts an image and audio recording into a lifelike talking-avatar video. Ready-to-use REST inference API, best performance, no coldstarts, affordable pricing.