An AI-generated human voice clone narrates an audiobook

• AI is entering the publishing industry in the form of artificial human voices narrating audiobooks.
• Some voice actors avoid jobs where their voices could be duplicated by AI.
• Industry experts told CNBC that AI is still far from replicating human narrators perfectly.

Audiobook narration may soon be done by artificial intelligence rather than voice actors, as technological advances have caused professionals across the industry to question their future.

Sales in the audiobook segment of the publishing industry are forecast to grow 26% annually from 2022 to 2030, reaching $35 billion by 2030, according to data from research firm Words Rated.

According to CNBC, as the voice acting industry is booming, voice actors are becoming more wary of the potential use of AI to replicate their voices.

AI-generated voices are already being used in several cases on Google Play and Apple Books. According to insider Katherine Tangalakis-Lippert, Apple has secretly released dozens of audiobooks on its Books app, narrated by voices named “Jackson” and “Madison.”

According to CNBC, Google Play offers publishers an auto-narration option for audiobooks, as long as the publisher owns the rights to the audiobook.

Voice actor Brad Ziffer told CNBC, “I refuse to take the job of taking my voice and creating an AI model.” “The best way to protect yourself is to just keep your distance.”

But experts say seamlessly replicating the way humans converse with AI is still a long way off. Humans exhibit unique intonations, rhythms and emotions when speaking.

“People are very sensitive to sound,” Narration Marketplace executive David Chicarelli told CNBC.

“The best thing humans can do that AI can’t do is timing,” Chicarelli said. “Whether it’s between awkward moments or a hilarious sense of comedy-like timing, it’s hard for AI voices to get this right right away.”

In a test of five audiobook excerpts in January, Insider was able to accurately determine which clips were narrated by humans and which were by Apple’s AI-generated voice. . But non-human voices weren’t entirely offensive.

“I’m very cautious about the world of AI. I believe AI has great potential…but it’s easy to exploit,” Ziffer told CNBC. “At the moment, I still believe there is nothing comparable to his VO in a real human being. No synthetic speech algorithm has yet been established that can perfectly reproduce all the nuances of the human voice.”