This web application is an intelligent music-recommendation platform that analyzes the semantic meaning and emotional tone of song lyrics to understand each user’s unique playlist “vibe.” By integrating Spotify authentication, the app imports a user’s playlists, retrieves lyrics through Genius, and transforms each song into a high-dimensional lyric-embedding vector that captures mood and thematic content across multiple languages.

It then computes playlist-level mood profiles, detects dominant emotional themes, and recommends new music by comparing these vectors to a dynamic candidate pool, including the Billboard Hot 100 chart. The system stores embeddings, lyrics, and metadata in a structured SQL database, enabling fast similarity search and mood classification. Overall, the app delivers highly personalized, mood-aware song recommendations based entirely on lyrical meaning rather than genre, popularity, or audio features.

• Frontend: HTML + CSS
• Backend: Python (Flask Framework)
• Relational Database: SQLAlchemy
• Tables: User, Playlist, Song, Artist, Billboard_entry2, Song_point, Artist_Song, Playlist_Song

• Some APIs used to retrieve song’s feature is depreciated
• In development mode, only 25 Spotify users are allowesd