Spotify to use Google AI to improve podcast, audiobook recommendations
Spotify has teamed up with Google Cloud to improve content discovery and to provide tailored recommendations for finding podcasts and audiobooks. It will employ large language models (LLMs) to identify a user's listening patterns. The platform plans to examine nearly five million podcasts and 350,000 audiobooks in Spotify's content library, to refine search results and user experience. As Spotify's exclusive cloud provider, Google Cloud has been collaborating with the streaming giant on various services.
Augmenting metadata for improved search results
Spotify will leverage Google Cloud's AI tools, to "augment" metadata. Metadata encompasses details like titles, host or author names, show notes, and other information that appears in search results on podcast apps like Apple Podcasts or Spotify. It's not exactly understood what Spotify means by "augment" in this case. However, the overall idea is to make it simple to find relevant podcasts and audiobooks based on user preferences.
Identifying harmful content and providing a safer listening experience
Spotify also aims to detect "harmful" content using Google Cloud's AI tools. Spotify currently enforces guidelines for podcasters and musicians that prohibit several forms of "sensitive" and "dangerous" content. For that, it uses a combination of automated technology and human reviewers. Spotify is yet to provide specifics on how the AI tools will be integrated into this process.
Spotify was early to adopt AI for music recommendations
Spotify has been a pioneer in using AI for music recommendations and now intends to do the same for non-music formats including podcasts and audiobooks, to boost revenue. The company had previously expected significant financial gains from its expensive investment in podcasts and audiobooks. As of September 30, Spotify reported a 4% growth in monthly active users (MAUs), reaching 574 million.
