In January, iHeartMedia announced its plan to leverage generative AI technology to convert five to 10 existing shows into multiple languages by the end of the current quarter. However, the company has decided to postpone this initiative.
The new timeline for iHeartMedia involves launching these translated shows by the end of the first half of 2024, as indicated by a company spokesperson to Digiday.
CEO Bob Pittman of iHeartMedia mentioned that while progress in experimenting with AI translation tools has shown promise, it has not yet reached the desired level for a full rollout. Pittman stated during a fourth-quarter earnings call on Feb. 29 that although advancements in quality are progressing rapidly, the final decision to launch will depend on further evaluation, likely around the upcoming earnings calls.
Despite announcements from podcast platforms like iHeartMedia, Spotify, and PodcastOne regarding plans to explore AI-generated audio translations, the actual implementation remains limited. For example, Spotify initiated a pilot program in September with selected podcasters to test AI-generated voice translations, but no significant updates were provided during the company’s Q4 earnings call on Feb. 6.
The uncertainty surrounding the quality of AI translations has made agency executives cautious. An anonymous executive from a podcast ad agency expressed concerns about investing in ads for such content until the quality reaches high standards and attracts a substantial audience, criteria that have not been convincingly demonstrated yet.
While the potential advantages of using generative AI tools for translating podcasts are clear in terms of cost-effectiveness and audience reach, accurately adapting colloquial expressions and cultural nuances remains a challenge. The technology primarily focuses on replicating a podcast host’s voice and adjusting it to present the show’s script in a different language.
Despite the challenges, some podcast networks are moving forward with launching translated shows to assess audience reception. For instance, PodcastOne is working with AI translation company Rask AI to translate selected podcasts into Spanish, starting with the true crime series “Bad Bad Thing.” The emphasis is on ensuring translation accuracy under the oversight of PodcastOne’s talent and production team.
During the Hot Pod Summit event in New York City, The Verge’s editor-in-chief Nilay Patel conducted a live test of AI-generated audio quality, revealing mixed results in the audience’s ability to differentiate between AI-generated and human-generated clips. Industry executives view the potential for global expansion of podcast shows using AI as a viable and cost-effective strategy, despite current quality concerns.
While some podcast hosts advocate for traditional translation methods involving human translators for quality assurance, others recognize the cost-effective appeal of AI tools for international expansion. However, the practical implications of integrating AI translations, including contract renegotiations with hosts, present additional considerations for podcast networks moving forward.
