Генерация видео с говорящим AI-аватаром

★ 7.7 · content

talking-head-production is a Claude Code skill that automates AI talking-head video creation via the inference.sh CLI. The recommended model is P-Video-Avatar — 18× faster and 6× cheaper than alternatives at $0.025 per second, with built-in TTS covering 30 voices across 10 languages. The skill also supports OmniHuman 1.5 and 1.0 for multi-character scenes with gestures, PixVerse Lipsync, and Fabric 1.0, all accepting either a text script or a custom audio file. Documented workflows include portrait requirements, style control via `voice_prompt` and `video_prompt`, full portrait-to-video pipelines using `pruna/p-image`, and two-character dialogue with separate TTS. Built for spokesperson videos, online courses, social media clips, presentations, and product demos.