How iHeartMedia kept the tone and personality of its AI-translated podcasts

The audio media giant has rolled out out AI-translated versions of 10 of its popular podcast shows, in six different languages – and in the original hosts’ own voices, thanks to voice-cloning technology. 

“We really wanted to make sure we didn’t launch until the technology was just right and we felt really good about putting these shows out in the world and doing it for a range of shows,” said Will Pearson, president of iHeartMedia’s podcast arm. “The technology has just improved so much in the last nine to 12 months.”

And naturally, media buyers need to be convinced too. 

The AI translations faced two main challenges: ensuring accuracy and preserving each show’s unique tone, Pearson said. The shows are not re-recorded by native speakers. iHeartMedia worked with AI audio and video company Speechlab to clone the voices of its podcast hosts and use them to read the show’s transcript in multiple languages. This allows listeners in international markets to enjoy these shows in their native languages – with the original podcast hosts’ voices.

iHeartMedia tested several AI voice companies before partnering with Speechlab, and secured buy-in from podcast hosts to launch the translated shows. That was the easy part, according to Pearson.

The harder part was getting the technology and process up to par. Speechlab cloned those hosts’ voices from snippets of their shows, and refined the audio to translate it into different languages while preserving the original tone and personality. These audio clips were shared with the podcast teams at iHeartMedia, who went back and forth between the podcast shows’ teams and with the talent and human teams of native speakers to review the AI-generated audio snippets. 

The podcast network would share feedback with the Speechlab’s team. For example, one host who has a slight accent had some words mistranslated. iHeartMedia also tested out the translations with groups of listeners who were native speakers.

Once this process was ironed out, it took “a matter of days” to translate a wider swath of podcast episodes, according to Pearson. About 15 episodes per show will roll out over the course of the next several weeks.

“The technology has gotten to the point where it’s really coming back very strong, even on the first pass,” he said. Podcast hosts vetted the AI-translated episodes, he noted.

“If you listened to one of these [shows] 12 to 18 months ago, it might have been accurate but not really in the sense of what a translator would have done. To make something feel conversational, it shouldn’t be one-to-one. There’s a nuance of language… you want the conversational flow to feel right – more for the chat shows than the narrative style shows. When Jay Shetty delivers his content, he has his own delivery,” Pearson said. “We want to make sure it captures that essence as much as possible.”

About a quarter of iHeartMedia’s monthly podcast downloads come from outside of the U.S., according to Pearson. “That’s not an insignificant amount of listenership,” Pearson said. “We do sell internationally. This is the early stages of getting new shows out in the world and getting to monetize from there.” iHeartMedia had about 555.6 million global podcast streams and downloads in May 2025, according to Podtrac data.

The AI audio translation technology isn’t replacing human translators at iHeartMedia. Pearson said it would have been “cost-prohibitive” to translate these podcast shows with translation editors. Pearson declined to share how much it cost to get these AI-translated versions of the podcast shows up and running. iHeartMedia worked with about two dozen internal and external people to help with the tech, edits and translations, he noted.

Kristen Coseo, director of podcast and digital audio strategy at Ocean Media, is skeptical of the current state of the AI technology, but lauded the iHeartMedia initiative.

“The ability to deliver content in language while preserving the host’s voice could maintain brand consistency and enhance listener connection, making it a smart strategy,” she said. “While AI voice cloning has significantly improved, there’s still a risk of sounding robotic or missing emotional nuance. If the quality is high, this could be a game-changer for reaching new markets.”

Coseo said she’d consider buying ads in the translated podcast episodes if there was data showing strong listener engagement and retention in international markets. “The opportunity to target diverse audiences with tailored content is compelling, but it hinges on the translations being seamless and culturally resonant,” Coseo said. “Authenticity is critical in podcasting.”

Another agency media buyer who requested to speak anonymously as they hadn’t heard the AI translated episodes — said while the idea and concept are good, they would need to consider the authenticity of the show’s emotion, context and rhythm before they bought ads in AI-translated podcasts. “Just because the content can be translated via AI doesn’t mean it should be,” they said. “We all know that word-for-word things do not translate exactly. So, I would be cautious and want to bring in someone that speaks the language to make sure it sounds the way it is supposed to, both from a content and advertising perspective.”

Media companies are finding more ways to use AI audio technology. Audio companies like Spotify and PodcastOne are experimenting with AI-translated podcasts as well. News publishers are also increasingly using AI to create audio products. Time and Business Insider introduced AI-generated audio news briefings this month.

iHeartMedia has disclaimers in its episodes and show descriptions that the podcasts are translated using AI technology. The episodes will be published weekly. iHeartMedia will analyze which podcasts do well in different markets, and evolve the strategy from there, Pearson said.

“I don’t know how these things will perform. But we do feel quite bullish on the growth of [this]. This is cost-effective enough that this is worth the investment to test this out,” he said.

The translated shows include “On Purpose with Jay Shetty”, “Revisionist History with Malcolm Gladwell”, “Stuff You Missed in History Class”, Stuff They Don’t Want You to Know”, “Before Breakfast with Laura Vanderkam”, “How to Money”, “Stuff to Blow Your Mind”, “Betrayal”, “The Girlfriends” and “Murder 101”. The shows will first come out in Spanish, then French, Arabic, Portuguese, Hindi and Mandarin, with plans to expand to even more shows and languages in the future.