The company states that it is trialing Google’s Internet of Things (IoT) tools to enhance the discovery of content and personalized recommendations for the mentioned audio.
Authored by Amrita Khalid, a prominent figure in the music industry newsletter Hot Pod, Khalid has delved into software, cybersecurity legislation, consumer electronics, and virtual realms for more than a decade.
It has become a common observation that podcasts and, to a lesser extent, audiobooks face challenges in discovery. In an effort to address this issue, Spotify has revealed its utilization of Google Cloud’s artificial intelligence (AI) tools to enhance content discovery and personalized recommendations for its audio offerings.
In this recent initiative, Spotify is leveraging Google Cloud’s Large Language Models (LLMs) to analyze its vast content library comprising approximately 5 million podcasts and 350,000 audiobooks. The objective is to “enhance” the metadata, as stated in a press release. The metadata of a radio show or audiobook typically includes information such as the title, host or artist name, show notes, and other relevant details that influence search results on audio platforms like Apple Podcasts and Spotify.
Google Cloud stands as Spotify’s exclusive cloud service provider, with collaborative efforts dating back to 2021, as reported by Protocol.
An additional goal of Spotify’s Google Cloud AI initiative seems to involve identifying and addressing “harmful” content, although specific details on this aspect remain undisclosed. The platform’s guidelines for content creators and consumers prohibit various forms of “sensitive” and “unsafe” content, which are now enforced through a combination of automated systems and human moderators. Further insights are awaited from Music, as reported by The Verge.
Moreover, Google Cloud’s AI tools will play a role in refining Spotify’s personalized recommendations for podcasts and audiobooks. By leveraging LLMs, Spotify aims to gain deeper insights into the content preferences of users, potentially leading to more accurate personalized suggestions.
As a result of these advancements, Spotify users may receive recommendations for new podcasts and specific episodes directly on their home screens. Additionally, a “More like this” section will offer similar content suggestions on each show’s page. However, the effectiveness of these features may vary, as highlighted by some users’ experiences. For instance, Spotify’s recommendations for the podcast “Overheard at National Geographic” included content on nursing, bipolar disorder, and true crime podcasts.
Earlier this year, Spotify introduced AI DJ, a tool focused on creating unique playlists using OpenAI’s relational AI technology. Feedback on this feature has been mixed. While AI DJ is tailored for music, Spotify’s integration of AI for spoken audio underscores its dedication to enhancing the user experience across its diverse range of products.
