Add a portrait and a voice track, then generate.

The result — a lip-synced talking-head MP4 — will preview here.