Photorealistic 3D/2D clone and realtime personalization at scale

Hi there, we aim to create a marketing campaign featuring a digital twin—a photorealistic 2D/3D clone of a celebrity—that vocalizes messages typed in by users. Imagine a landing page with a form; based on the text entered by users, the digital clone will either speak or, preferably, rap the message in a simple manner. While the concept is similar to the campaign found at Messi Messages, we believe we can execute it in a manner that surpasses the current quality. Is this possible with Unreal Engine?