Quincy Surasmith is a radio journalist and actor, but you may also hear his voice and never realize it. That’s because he’s been the voice of Thai-speaking cartoons, chattering background crowds, and characters without major speaking roles. It’s not all glamorous. “I’m making grunting noises, getting beat up by some guy,” Suarasmith says. “It takes specific improv and acting skills.”
Soon those grunting and background chatter performances could be at risk of being replaced by artificial intelligence. Voice acting is a highly specialized skill, but generative AI is becoming more adept at talking back, from cloning celebrity voices to narrating audiobooks. The tech doesn’t just create more competition for jobs; voice actors also worry about their vocals being stolen and copied to promote mis- and disinformation, becoming victims of deepfakes, or hearing themselves appear in pornographic content without their consent—all situations that would damage their professional reputations and plunder their biggest, most recognizable asset: their voices.
Industry experts agree that some jobs will be lost in the gen-AI boom. Cheap, entry-level voice work can likely be replaced by machine-generated vocals. But they’re also optimistic that AI can’t fully automate what voice actors do. To get the right emotion, dialects, and artistry behind the craft, producers will still need to hire humans. For animated characters in high-production-value shows, having human actors to convey cultural nuances is vital. But Surasmith worries that AI may be cheaper to hire for some of the smaller gigs: “Is that something production companies will think, ‘Hey, that’s the replaceable part?’”
AI tends to make the voices “as boring as possible,” says Dan Lenard, president of the Word-Voices Organization, a nonprofit association for voice work. The technology could be a low-cost fix for companies that make, say, informational HR videos, but synthetic voices don’t engage people in the same way as humans do. “Every voice is different, every accent is different, and I think that’s one of the things AI cannot duplicate,” Lenard says.
Still, companies are eying the opportunities. Last week, Spotify announced a pilot for a translation feature for podcasts. It’s powered in part by OpenAI’s generative voice tech and translates podcasters’ voices into other languages. The first batch features popular figures like actors Dax Shepard and Monica Padman, sportscaster Bill Simmons, and former Daily Show host Trevor Noah. Then, OpenAI also announced it had integrated voice tech into its chatbot ChatGPT, so people can speak back and forth with it.
The rapid advances in tech threaten more than just voice artists’ jobs; the actors also worry that their voices could be used to create new content they haven’t signed off on. Two years ago, the team behind Roadrunner, a documentary about the late Anthony Bourdain, used AI to clone his voice and have it read an email he had written. The move set off alarm bells in Hollywood and raised ethical questions about how AI might bring people’s voices, gestures, and words back to life after they’d died.