Spotify launches Mixes, using machine learning to personalise playlists
It draws data from the 500 billion actions that happen every day on the streaming platform.
Image: Spotify
Spotify has launched Mixes, which are personalised playlists that have been curated with machine learning to recommend users new music based on their listening history.
The feature launches today (5 April), and can be found under Search in the platform’s Made For You hub. The playlists arrive in three forms: Artist Mix, Genre Mix and Decade Mix. Each begins with songs that the user had already liked, then recommends unheard tunes.
According to a statement, Spotify Mixes were inspired by one of its most popular playlists: Daily Mix. However, these new playlists are deeply personalised to an individual listener’s tastes with the aid of machine learning.
Gustav Söderström, Spotify’s Chief R&D Officer, elaborated on Mixes during the brand’s Stream On event in February.
“There isn’t just one Spotify experience. There are actually more like 345 million different Spotify experiences – one for each listener,” he said. “Every day, half a trillion events – whether they are searches, listens, likes […] take place on Spotify, powering and guiding our machine learning system. This gives us the ability to drive discovery in a way that audio has never seen before.”
If you’re interested in learning more about Spotify’s machine learning practices, Söderström dove deeper into the subject in the latest instalment of Spotify: A Product Story, a podcast series about the streaming giant’s product strategy.
For more music technology news, click here.
