Spotify’s Mind-Blowing New Thing

We’d all like to speak multiple languages, right? I quite fancy learning Japanese, for instance. Well, now I’ll never need to because Spotify can pretend that I already do.

This week, Daniel Ek – founder of Spotify and generally very impressive chap – announced that they’re testing a new trick that could genuinely change how podcasts are delivered around the world. We’re frequently trying to crack brand podcast briefs intended for a global audience, but that almost inevitably means recording them in English. Because everyone out there speaks English, right? Well, no. And certain countries are more resistant to this than others. For instance, Spanish podcast audiences are particularly resistant to listening to anything that isn’t in Spanish. Podcasting is in danger of becoming another medium where the biggest content requires an element of language-colonialism.

But localising unscripted podcasts is really tricky, and most of the most popular podcasts are unscripted. Do you have episodes with actors speaking the words in different languages? Just provide a transcript? Re-record with a completely new set of people (which we’ve done before)? With the world’s fastest growing markets including the Middle East, South Asia and South America, this is a frustrating and costly process.

Well, as with so many aspects of modern life, the answer is AI. Because Spotify’s technology can … stick with me … sample the speaker’s voice, understand what they’re saying, translate their words into another language, and then replace the audio with them speaking the foreign language in their own voice. Unbelievable.

So now, in Spotify’s test, listeners who enjoy celebrities being asked increasingly personal questions until they cry will be able to hear Stephen Bartlett’s Diary of a CEO and select the language they’d like him and his guests to speak (and cry) in. Not dubbed, but in their own voice. I know I can sometimes be cynical on this newsletter, but I genuinely think this is the most game-changing piece of tech I’ve seen in the podcasting world for years. In the future, we can make one podcast, recorded in a single language, but available to anyone in their local tongue. Sorry, I can’t stop going on about it.