Generate lip‑synced video from a face image and audio