A couple of months ago Spotify launched Voice Translation; a tool that enables podcasters to translate their spoken content into a language of their choice. Spotify.
Using OpenAI’s voice generation technology to match the speaker’s original style, so that it sounds more authentic than classing dubbing. There will no doubt be limits to the authenticity of this content, as specific cultural inferences or accents might literally get lost in translation through AI.
For the pilot, Spotify worked with a select group of podcasters on specific episodes of their podcasts. As a result, Spotify Premium listeners can now listen to episodes from Steven Bartlett’s “The Diary of a CEO” podcast in Spanish. Spotify’s voice generation effort is one of the efforts led by Ziad Sultan, Spotify’s VP of Persona Personalisation, utilising AI to personalise the content it offers to its users (Spotify DJ is another good example here).
Main learning point: I wonder about both the podcaster and the listener being able to check the authenticity of the content. How will content creators be able to check the quality of their translated content?
Related links for further learning:
