Give a voice sample and a starting frame — the scenario clones the voice, prepares the description and assembles a talking-avatar video.