Emotion-Adaptive Music Recommendation System Using LLMs and Real-Time Sentiment Analysis

Abstract

Music emotionally affects individuals, personalized music suggestions are required for an enhanced listening experience. This work introduces an Emotion-Adaptive Music Recommendation System leveraging large language models (LLMs) and sentiment analysis to recommend personalized songs based on user emotions. The method guarantees an organized emotional categorization of music by pre-labeling every song with moods from the GoEmotions database following processing of a Spotify dataset. In a bid to interpret the identified sentiment to the preprocessed data for song popularity ranking, a RoBERTa chatbot interacts with users to capture and assess the mood inputs. To provide the user preferences for specific artists and monitor popularity priority, additional filtering is utilized.To enable smooth music playback, the Spotify API is used to build a simple and user-friendly interface. This work aims to improve emotional connection and listening enjoyment by combining deep learning, natural language processing (NLP), and recommendation techniques.