Spotify's new AI 'DJ' will talk you through its recommendations

Generative AI is absolutelyeverywhererightnow, so it’s no surprise to see Spotify putting it to use in its latest feature, simply called “DJ.” It’s a new way to immediately start a personalized selection of music playing that combines Spotify’s well-known personalization tools that you can find in playlists like Discover Weekly as well as the content that populates your home screen with some AI tricks. I got early access to DJ and have been playing with it for the last day to see how Spotify’s latest take on personalized music works, but the feature is available as of today in beta for all premium subscribers in the US and Canada.

While Spotify has loads of personalized playlists for users, I’ve found that the app lacks a simple way to tell it to just play some music you like. On Apple Music, for example, I can ask Siri to play music I like and it’ll start a personalized radio station based on music I’ve played alongside some things it thinks I’ll enjoy but haven’t played before. It’s a reliable way to jump right into my collection. In the same vein, Spotify’s DJ pulls together a mix of songs you’re currently listening to, old favorites you might have forgotten, and new tunes that fit in with what it thinks you’ll like.

The AI twist to DJ comes in the form of a literal DJ, which speaks to you in an AI voice generated by Sonantic, a startup that Spotify bought last year with a focus on generating realistic speech. In this case, the DJ’s voice model was trained on the voice of a real human, Spotify’s own Head of Cultural Partnerships, Xavier “X” Jernigan. Jernigan hosted “The Get Up,” Spotify’s morning show that combined recorded segments with music tailored to your tastes.

The DJ’s voice is generated through AI, and so are the things it says to you. When you first kick off a DJ session, you’ll get a quick overview of what you might expect to hear. For example, the first time I started up DJ, “X” came on and told me that it was a DJ designed for music and that it knew what I liked and for starters it was going to play me some Jenny Lewis. Sure enough, Lewis’s “Do Si Do” kicked things off, along with a few other songs with a similar vibe. At the top of the now playing screen, you’ll see a little info on how the song was picked, like “based on recent listening,” “throwbacks,” “recommended for you” or “from your past.”

Once you start a segment, you’ll generally hear a handful of songs that fit into the category, but if you want to change things up you can just tap the DJ button in the lower right corner of the now playing screen. At that point, X the DJ pops back up to give you some info about what’s coming up next. When I just tapped it, X said, “OK, changing it up. Here are our editor’s picks for the best in hard rock this week, starting with Motionless In White.”

Spotify says that none of the dialog you hear from X is pre-recorded; it’s all generated on the fly by OpenAI. However, the company wanted to make it clear that it looks at generative AI as a tool for its music editors, not something that it is just trusting to get everything right. Spotify’s VP of personalization Ziad Sultan told Engadget in a product demo that the company put together a “writers room” of script writers, music editors, data curators and engineers, all of whom are working together to make sure that the bits of info that the AI DJ drops are useful, accurate and relevant to the music you’re hearing.

Sultan stressed that Spotify’s usage was a lot different than implementations like free-form text, image generation and other such AI use cases. “We’ve built a very specific use case, and we’ve made a few choices about how it’ll be implemented,” he said. “The most important one is the creation of that writer’s room – we’re taking this [AI] tool and putting it into the hands of music experts.”

What’ll make Spotify’s DJ work or fail is whether it can pull up music you want to hear. From that perspective, Spotify isn’t doing anything wildly different than it already does: analyzing your listening history and finding stuff it knows you like and things it thinks you’ll enjoy. And as with everything else you do on Spotify, your DJ usage will be analyzed so that it can get better at serving you tunes you want to listen to. At the beginning, anyway, the AI DJ aspects are being used as small augmentations to a personalized music channel – and as long as Spotify can continue to know what songs you love and which ones you’re likely to fall in love with, DJ should be a useful addition.

from Engadget is a web magazine with obsessive daily coverage of everything new in gadgets and consumer electronics https://ift.tt/HWSscv2
Nathan Ingraham

Generative AI is absolutelyeverywhererightnow, so it’s no surprise to see Spotify putting it to use in its latest feature, simply called “DJ.” It’s a new way to immediately start a personalized selection of music playing that combines Spotify’s well-known personalization tools that you can find in playlists like Discover Weekly as well as the content that populates your home screen with some AI tricks. I got early access to DJ and have been playing with it for the last day to see how Spotify’s latest take on personalized music works, but the feature is available as of today in beta for all premium subscribers in the US and Canada.

While Spotify has loads of personalized playlists for users, I’ve found that the app lacks a simple way to tell it to just play some music you like. On Apple Music, for example, I can ask Siri to play music I like and it’ll start a personalized radio station based on music I’ve played alongside some things it thinks I’ll enjoy but haven’t played before. It’s a reliable way to jump right into my collection. In the same vein, Spotify’s DJ pulls together a mix of songs you’re currently listening to, old favorites you might have forgotten, and new tunes that fit in with what it thinks you’ll like.

The AI twist to DJ comes in the form of a literal DJ, which speaks to you in an AI voice generated by Sonantic, a startup that Spotify bought last year with a focus on generating realistic speech. In this case, the DJ’s voice model was trained on the voice of a real human, Spotify’s own Head of Cultural Partnerships, Xavier “X” Jernigan. Jernigan hosted “The Get Up,” Spotify’s morning show that combined recorded segments with music tailored to your tastes.

The DJ’s voice is generated through AI, and so are the things it says to you. When you first kick off a DJ session, you’ll get a quick overview of what you might expect to hear. For example, the first time I started up DJ, “X” came on and told me that it was a DJ designed for music and that it knew what I liked and for starters it was going to play me some Jenny Lewis. Sure enough, Lewis’s “Do Si Do” kicked things off, along with a few other songs with a similar vibe. At the top of the now playing screen, you’ll see a little info on how the song was picked, like “based on recent listening,” “throwbacks,” “recommended for you” or “from your past.”

Once you start a segment, you’ll generally hear a handful of songs that fit into the category, but if you want to change things up you can just tap the DJ button in the lower right corner of the now playing screen. At that point, X the DJ pops back up to give you some info about what’s coming up next. When I just tapped it, X said, “OK, changing it up. Here are our editor’s picks for the best in hard rock this week, starting with Motionless In White.”

Spotify says that none of the dialog you hear from X is pre-recorded; it’s all generated on the fly by OpenAI. However, the company wanted to make it clear that it looks at generative AI as a tool for its music editors, not something that it is just trusting to get everything right. Spotify’s VP of personalization Ziad Sultan told Engadget in a product demo that the company put together a “writers room” of script writers, music editors, data curators and engineers, all of whom are working together to make sure that the bits of info that the AI DJ drops are useful, accurate and relevant to the music you’re hearing.

Sultan stressed that Spotify’s usage was a lot different than implementations like free-form text, image generation and other such AI use cases. “We’ve built a very specific use case, and we’ve made a few choices about how it’ll be implemented,” he said. “The most important one is the creation of that writer’s room – we’re taking this [AI] tool and putting it into the hands of music experts.”

What’ll make Spotify’s DJ work or fail is whether it can pull up music you want to hear. From that perspective, Spotify isn’t doing anything wildly different than it already does: analyzing your listening history and finding stuff it knows you like and things it thinks you’ll enjoy. And as with everything else you do on Spotify, your DJ usage will be analyzed so that it can get better at serving you tunes you want to listen to. At the beginning, anyway, the AI DJ aspects are being used as small augmentations to a personalized music channel – and as long as Spotify can continue to know what songs you love and which ones you’re likely to fall in love with, DJ should be a useful addition.

https://ift.tt/hzv5iIt February 22, 2023 at 04:00PM