Meta AI Gets Photorealistic Stand-ins and Natural Voice Chats

Why Trust Techopedia
Key Takeaways

  • Meta AI is getting virtual personas that can stand in for you.
  • They'll be available in early 2025.
  • Natural voice conversations with Meta AI are available today.

Natural Voice Chats, photorealistic AI avatars, and automatic AI video dubbing and lip syncing were just a few of the features we saw unveiled at Meta Connect.

From today, Meta is adding natural voice interactions to Meta AI, so you can have natural AI conversations across Meta apps including Messenger and Instagram.

Zuckerberg said that “voice will become one of if not the most frequent ways we interact with AI.”

As well as system voices, Meta confirmed some iconic voices coming to Meta AI, including Judi Dench, Kristen Bell, John Cena, Keegan Michael Key, and Awkwafina.

We saw a live demo of Zuckerberg talking to AI Awkwafina, asking if live demos are risky. The response sounded pretty natural and conversational. AI Awkwafina responded that yes, live demos were risky, and that they can be “unpredictable, prone to technical issues, and potentially embarrassing.”

AI Studio’s Photorealistic AI Creator Avatars Look and Act Just Like You

Meta AI Photorealistic Avatar Stand-in
The photorealistic personas in Meta AI. | Source: Meta

Next up, we got a peek at a new functionality for creators, coming early next year in the US. With AI Studio, creators can make their own photorealistic AI avatar to engage with their audience. Meta has already started releasing a text-only version, but a more interactive version is in the works.

Dan Allen Stevenson III, author of the first text-only creator AIs, came on stage for a live demo where Zuckerberg asked his avatar a variety of relevant – and not so relevant – questions, all of which received a realistic response.

The photorealistic avatars look just like their creators and can send and reply to DMs, share links, showcase past videos, and guide people through conversations. They’re great for creators who receive a ton of DMs, saving them time responding, and can even answer FAQs.

Creators also have a lot of control over how their avatar responds. The AI avatars are trained on the way creators write, from content posted on their Threads, Instagram, and Facebook.

In the live demo, Stevenson’s avatar looked and sounded eerily like the author, answering a variety of questions posed by Zuckberg.

AI Automatic Video dubbing and Lip Syncing is Creepy but Cool

Finally, we heard about how Meta is experimenting with AI automatic video dubbing on Reels. This feature will allow creators to reach more people worldwide, in a variety of languages.

Meta is starting with English and Spanish and will roll out more languages in future.

We saw a couple of Reels of creators speaking in Spanish and English, which were then automatically dubbed into either English or Spanish, as required, using the new AI feature.

Automatic lip syncing ensures the creators’ lips are synced with the translation, so it looks like they’re really speaking another language; this is a little creepy, but very cool. This will help creators engage with audiences who speak other languages.