How does Spotify Know You So Well?
Spotify has become one of the largest music streaming services and much of their success is due to their specifically tailored playlists. Not only does Spotify allow users to build their own playlists, but Spotify connects users with their friends and provides new music suggestions based on three recommendation methods that use social and music network connections and clustering. The three methods that Spotify uses are Collaborative Filtering, Natural Langrage Processing, and Audio Models.
Collaborative Filtering is an algorithm that relies on implicit user feedback. It relies on streaming counts and data, and user visits to artist’s pages. This method of recommendation utilizes a listener’s connection to other listeners and uses that connection to provide song recommendations that similar listeners already like. This data is stored in User Vectors, that store the music preferences of a listener, and Song Vectors, that store a song’s music profile. These vectors are compared and used to recommend similar song to similar users from Spotify’s massive database of more than 300 million songs. This method works due to the connectivity of users and songs in the database.
The Natural Language Processing component of the recommendation algorithm uses text from blog posts and music articles to connect songs and artists to each other. The algorithm sees what artists are talked about in clusters and recommends similar artists to listeners that show interest in one artist of that cluster. This algorithm takes advantage of the tight-knit network of musical artists and puts them into cultural vectors to recommend similar content.
The third recommendation method that Spotify uses Audio Models. Collaborative Filtering and Natural Language Processing are useful to connect listeners with music that others are already listening to and are talking about. Audio Models are useful for analyzing raw audio data and recommending new songs that have not become popular yet. This algorithm uses Convolutional Neural Networks that use clustering to identify similarities in time signature, key, mode, tempo, and loudness of audio tracks.
These three models are stored and used with clustering mechanisms to provide users with song recommendations and individually targeted playlists, like your “Discover Weekly”. Spotify’s use of clustering and networks to provide users with an individualized listening experience has contributed to Spotify’s success.