Spotify is testing a new AI-based feature that will allow you to listen to any podcast from around the world in Spanish or English, even if the presenters are Chinese or Arabic.
Voice Translation is a new feature launching today (September 25th) that will allow you to listen to certain podcast episodes in a different language, but with the original speaker’s voice… Yes, this might sound familiar to you.
The tool, developed by Spotify with the assistance of OpenAI’s Automatic Speech Recognition (ASR) system, Whisper, uses a generative AI voice-to-text model to translate audio files and a voice replication model to match the style of the original speaker.
Spotify will begin with its own podcasts
Among the first presenters to be part of this new feature are Dax Shepard, Monica Padman, Lex Fridman, Bill Simmons, and Steven Bartlett. In other words, some of the prominent and important shows on the platform.
However, not all episodes of their respective podcasts will be available in multiple languages right away. Instead, keep an eye on the episode “Entrevista con Yuval Noah Harari” in Lex Fridman’s podcast, “Kristen Bell, por la gracia de Dios, vuelve” in Armchair Expert, and “Entrevista con la Dra. Mindy Pelz” in The Diary of a CEO with Steven Bartlett, which are already available in Spanish.
Spotify states that more episodes will be available in the coming days and weeks, with translations into French and German. You can find them in your mobile or desktop app’s Now Playing view, and more voice-translated episodes will be added to a dedicated Voice Translations hub.
Indeed, AI is going to be truly game-changing now
The top music streaming services quickly jumped on the generative AI bandwagon following the buzz generated by OpenAI’s ChatGPT.
While they were already using machine learning to identify patterns and trends in music listening habits and better recommend new songs (e.g., the Discover Weekly playlist), new ways of utilizing this technology have been introduced.
Spotify’s AI DJ, which uses AI-generated voice to recommend new songs, is one of them. There’s also the Universal Music partnership with Endel to create ambient sounds like forest noises and flowing water using AI, as well as numerous music generators, including those from Meta and Google.
But undoubtedly, what was most eerie was the idea of using generative AI to create podcasts from scratch. Several generative AI podcasts emerged from experimentation in this field, including The Joe Rogan AI Experience and Hackers News Recap, to name a couple.
While the idea of creating podcasts with AI is somewhat unsettling, using generative AI to translate podcasts is the kind of use that many would support, as it makes them more accessible to all listeners, regardless of the language they speak.