European music streaming service Deezer has turned to artificial intelligence to help users create playlists. The new Playlist with AI feature produces a list of songs based on text prompts that describe anything from a music genre or time period to your current emotions or activities.
Playlist with AI is currently in beta, with 5% of Deezer subscribers randomly selected to test the feature. However, the company is likely keen to roll it out to more customers, as the idea of an AI tool that turns text into playlists is spreading rapidly. Spotify’s AI DJ, Amazon Music’s new Maestro, and YouTube Music’s AI-generated radio are all already available or in their own testing phase. Like its rivals, Playlist with AI interprets text prompts provided by users using an AI model to select suitable songs from Deezer’s music library, including ones the user has never played before.
“We are excited to offer this AI-powered feature to Deezer users around the world,” said Deezer VP of Product Alexandra Leloup in a statement. “Whether you need the perfect soundtrack for a workout, a romantic evening or a nostalgic trip down memory lane, our Playlist with AI feature will create a new music experience in seconds and offers endless possibilities to easily discover new music.”
DJ AI
Deezer has been toying with AI for a while now. Users can already use the Flow feature to compile AI-generated playlists, though these are limited to specific moods, genres, and songs previously added to a favorites list. There’s also the AI-powered song identification tool SongCatcher.
AI-powered playlist creators will only increase the rivalry between platforms that offer them. As personalized playlists become more popular, streaming services may treat them as a standard and necessary feature. There are broader implications for the music industry, too. AI-powered recommendations could help emerging artists gain exposure. By including lesser-known tracks that fit user-specified criteria, AI can introduce new talent to an audience that might otherwise never discover them.
This is all well and good, except of course if labels pay to boost the weight of their clients’ music in algorithms so that it gets played more often in more possible text prompts — a kind of postmodern payola (the scandal when labels secretly paid radio stations to play certain songs). That could ruin playlist personalization and annoy music fans if not done openly and well-marked.
