I wonder if that would be a genuine use case for “AI”. If the voice actor consents to have his voice represented in such a scene but doesn’t want to play it out in a studio, the computer model could take over that part.