Spotify To Use AI To Replicate Podcasters’ Voices

Spotify on Monday announced a new artificial intelligence-powered feature that can translate podcasts into different languages using the host’s own voice.

CNBC reports the company said the feature relies on OpenAI’s voice generation technology, which was also announced Monday. The technology can create “realistic synthetic voices” from just a few seconds of speech, OpenAI said in a release. Spotify said it will help make the podcast listening experience more authentic and natural by maintaining podcasters’ “distinctive speech characteristics.”

The company worked with podcasters such as Dax Shepard, Lex Fridman, Monica Padman, Steven Bartlett and Bill Simmons on a select number of past and upcoming episodes. Spotify said the episodes were initially translated into Spanish, French and German, and they will be available to both free users and paid subscribers in the coming days and weeks, according to the release.

A spokesperson said these are some of the most popular languages for users who listen to podcasts on the platform, and that the pilot stage has focused first on Spanish and will “follow quickly” with German and French.

Spotify did not share specific details about the cost of the new feature with CNBC. A spokesperson added that the company is exploring opportunities to “bring this service to even more podcasters globally.” The company is already working on comedian Trevor Noah’s upcoming new podcast, for instance.