ElevenLabs Adds AI-Powered Celebrity Voices to Recent Digital Narrator — But Is It Unthreatening?

A week ago, ElevenLabsThe voice AI startup founded by former Google and Palantir engineers made headlines with its first major consumer-facing product, the Reader app.

Currently available on iOSThe product is a dedicated voice recording solution that converts any text file or link from the web into AI audio, read by a variety of AI voices and accents. Today, the company announced that it is expanding that library of voices in the app with the AI ​​voices of deceased Hollywood stars Judy Garland, James Dean, Burt Reynolds, and Sir Laurence Olivier.

The company has partnered with CMG Worldwide, a company that manages and protects the intellectual property of living and deceased celebrities, to recreate and launch the iconic voices, and plans to build on that work by introducing many more renowned AI voices in the coming months.

Reader gives AI a voice to any digital text

While ElevenLabs has focused specifically on the inventive industry, offering AI models for text-to-speech and speech-to-speech conversion, dubbing, and sound effects, the Reader app takes its text-to-speech research into a more tailored form. All a user has to do is feed in a link or file of any digital text—whether an article, PDF, newsletter, or 300-page e-book—and the app will instantly process the text and start AI voice narrating, with a green highlighter following and highlighting each word the AI ​​says.

The feature is available in English, although users can customize their experience by choosing from 11 voices and accents, ranging from male to female, American, Austrian, and British English. Now, Iconic Voices, launched today, enhance that experience by allowing users to discover and experience content in the voice of deceased celebrities.

Imagine being able to listen to L. Frank Baum’s The Wonderful Wizard of Oz with the voice of the overdue Judy Garland, who starred in the film adaptation of the novel.

For family members of deceased stars, AI-powered voice playback is an opportunity to ensure that the celebrity’s legacy lives on, that their current fans have a way to reconnect with them, and that new-age users have a way to discover them. For ElevenLabs, meanwhile, the announcement is expected to boost engagement on its fresh app.

“Judy Garland, James Dean, Burt Reynolds and Sir Laurence Olivier are some of the most iconic actors in history. We deeply respect their legacy and are honored to have their voices as part of our platform,” said Dustin Blank, Head of Partnerships at ElevenLabs. “Adding them to our growing roster of narrators is a major step forward in our mission to make content available in every language and voice.”

Are AI voices sheltered from abuse?

One of the biggest concerns with voice-cloning technology—like the one discussed here—is that voice reconstructions of renowned people can portray them as saying things they never said in real life. The Biden robocall incident is the biggest example of this problem. In the same way, what if a CEO’s voice were cloned to make him say things that could potentially ruin his reputation or the reputation of his company?

ElevenLabs says it understands these concerns and is taking steps to expand its partnership with the iconic voice feature, with a special focus on security.

Sam Sklar, who leads growth marketing at ElevenLabs, told VentureBeat that the company retains full control over the celebrity voices and makes them available exclusively in the Reader app, which is designed so that users can only convert the digital text into an AI narrative for individual exploit — rather than share or download it.

“For example, through the Reader App, you can select an article on VentureBeat and have Judy Garland narrate it just for you. You can’t access her voice through the ElevenLabs Voice Library (the company’s separate web product). That means they can’t be used in conjunction with our standard text-to-speech tools on the platform, nor can you download or share the content they narrate through the Reader App,” he explained.

If a user sends harmful content as text to record its iconic voice narration using an assistive device, the company won’t even generate an AI voice. It has placed automated and human moderation processes in between to identify and block hate speech and other forms of text that violate its terms of service.

As for the chances of the voice library being misused to clone celebrity voices from scratch, Sklar says the platform is built with several safeguards, including voice captcha verification that matches audio samples submitted for cloning with a recording of the user’s voice. If the voice doesn’t match after a few tries, the clone request isn’t processed. There’s also a “no go” policy for voices that prohibits cloning voices deemed high-risk.

“Any attempts to clone these votes will be blocked,” Sklar said.

While these steps do reduce the likelihood of cloning the voices of celebrities, actors, and business executives, there could still be instances of abuse. For example, malicious users could create content for the Reader app in a way that bypasses the company’s moderation measures.

In the long run, it will be fascinating to see how the iconic voice capabilities, which were introduced as a deal for fans and enthusiasts, affect the industry. The Reader app that supports them is rolling out globally and to Android devices this summer. Support for more languages ​​is also on the way.