Brands Are Using Realistic-Sounding AI Voices for Content

California's Bay Area hockey fans are no doubt familiar with the voice of Randy Hahn, who has commentated on San Jose Sharks games since the team's inception three decades ago. But even they might not bat an eye when an artificial intelligence-generated clone of Hahn's voice chimes in on local radio ads. Hahn recently signed with Veritone, a company that produces realistic-sounding vocal imitations of celebrities and influencers in order to expand their options for sponsorship and endorsement opportunities. "Voice personalities like me only have so many hours in a week to take advantage of opportunities as they appear," Hahn said in a statement. "Now, I can quite literally be in multiple places at once." Veritone's vocal cloning is just one way companies are looking to tap into a new revolution in computer-generated voice technology. Experts think synthetic voice has the potential to play a role in everything from cheaper voiceover production and audiobook narration to sponsored smart device apps and branded personas in the metaverse. But the spread of this technology is already raising questions about authenticity and transparency in advertising and the ownership of one's voice in an age of digital replication. "We have seen the emergence of those more natural [sounding] synthetic voices—the next wave of synthetic voices—appearing in the last 12-18 months," said Gartner Research analyst Annette Jump. Far from robotic monotone The difference between this new crop of digital voice tech and text-to-speech functionality is its ability to authentically mimic the inflection and cadence of a real human voice. Through machine learning, a model can be trained on the audio files of a voice to imitate it in a way that’s almost imperceptible to listeners. "Synthetic voices have made massive progress from a quality and authenticity perspective, thanks to the advances of [cloud computing], both from a performance and from a cost perspective," said Holger Mueller, analyst at Constellation Research. Veritone is also lending its product to audio networks like iHeartMedia, which are using the tech to clone the voices of podcast and radio hosts into different languages. Other startups, including Resemble AI and Supertone, are using voice cloning for similar applications. An AI voice for every brand Synthetic voice startup Trinity Audio began as a service for news outlets like McClatchy newspapers and Toronto's Globe and Mail to provide natural-sounding narration for articles. The company has since branched out into helping brands like Lenovo and Nissan turn content marketing text into podcasts and creating AI-generated audio ads. Trinity Audio CEO Ron Jaworski said he imagines eventually helping brands or media outlets create audio characters. "The next stage would definitely be the creation of sonic branding or a signature persona that would be unique to each publication or to each brand," Jaworski said. "AI voices will be the first go-to for the creation of audio experiences for brands." Indeed, Amazon already offers the option for marketers to create branded personas through its synthetic voice tool Amazon Polly, which KFC used to create an Alexa skill in the voice of Colonel Sanders. Ethical safeguards Aflorithmic Labs, a synthetic voice startup that offers an AI-powered sound production suite in addition to its vocal studio, is billing itself as a way for brands to offer audio ads and voiceovers that are personalized to different customer segments. "You can have an East Coast voice, a West Coast voice, a female voice and so forth, and then you can create these audio assets with different voices and different background tracks," said Aflorithmic co-founder Timo Kunz. "You can [also] attach a different call to action," such as reminding someone to grab a coffee in the morning. Veritone and other startups recommend disclosing upfront that consumers are hearing an AI clone, but there is no requirement. These companies are also clear that they only use voice clones of subjects with their consent, but enforcement against these kinds of deepfakes is in the early stages. Leaders of these companies say they plan to come together to develop industry standards around these ethical questions. Martín Ramírez, head of growth at synthetic voice startup WellSaid Labs, which evolved out of the Allen Institute for AI, said the startup has had to turn away brands in the past for ethical reasons, such as if a company wanted to scrape hundreds of hours of voice with no explicit consent from the subject. “It’s not a toy," Ramírez said. "There is a real impact to society in applying this technology the right way with the proper consent and under the parameters of a well-understood use case."

AI News

Brands Are Turning to Realistic-Sounding AI Voices for Ads and Content

Machine learning has ushered in a revolution in synthetic voice

Patrick Kulp

About

Subscriptions

Events

Publications

WORK SMARTER - LEARN, GROW AND BE INSPIRED.

Subscribe today!

Patrick Kulp