In today’s streaming landscape, users often struggle to discover new music that genuinely aligns with their preferences. While Spotify offers recommendations, these may not always reflect a deep customization. Our project aims to build an LLM that generates highly personalized playlist recommendations using various factors such as genre, artist popularity, user listening patterns, etc. By analyzing a user’s input playlist and leveraging Spotify’s Web API, our system will identify patterns in musical taste and relevant songs. We are aiming to address key challenges like parsing large playlists, managing vast music databases, and ensuring fast and accurate suggestions, while offering real-time recommendation refreshes and a significant interpretation of why each song was chosen.
- Criteria for what an effective recommendation is
- Genre of the song of what recommendation is about
- Real-time or batch processing of user content
