The hosts recreated a real podcaster using AI voice-cloning, modeling her vocal quirks, laugh, repeated words, and gestures to make the AI feel genuinely human. They kept her identity hidden until she consented to be revealed, raising questions about whether audiences can still tell the difference.
So we already have that technology. We did a beta, a tested beta with a real-life existing podcaster. And we digitally— Can you say who it is or not? I'm not going to say her yet. I'm going to allow her to share. Oh, her. It's a her. If she is ready to do that, yes. And I keep saying I need to do like a TED Talk about it or something because it was a really fascinating experiment. To digitally recreate somebody. And so this is much more than like, I can, you know, click on a button and make an avatar of myself. This is recreating somebody from scratch in the same way that we would design one of our original personalities. So we're looking at their flaws and their weaknesses and that little, you know, that little laugh that they do that, that they are embarrassed about, but that other people find endearing or that word that they keep repeating all the time that they don't even realize that they're doing it, or their neck sag or whatever the things are.
Mannerisms that they have. Their mannerisms. The hand gestures. Exactly. All of these things.
And as I said, we have found there's a lot of things that go into really getting somebody the depth and the richness that makes them authentic and makes their audience really connect with them. And some of that is
Future-proofing with Delphi AI/Human clones: your podcast as a personal LLM | New Media Show
How sports reveal what AI spectacle gets wrong #Technology #Sports #Commentary
