Emotion-Aware Music Recommendation System Using Chatbot for Mood Detection

Abstract

Problem: With the increasing concern around mental well-being, there is a growing need for systems that can understand user emotions in real time and provide comforting responses. This project introduces MindBeats, an emotion-aware music recommendation system that leverages a chatbot to interact with users, detect their emotions through text-based conversations, and offer personalized Spotify music recommendations to uplift their mood. Methodology: The system utilizes a hybrid deep learning model combining Convolutional Neural Networks (CNN) and Bidirectional GRU layers for emotion detection from user text inputs. The architecture is built using Python (Flask) for the backend and React for the frontend. Text preprocessing includes tokenization and padding. Follow-up questions are dynamically generated using the Gemini API based on detected emotions, enhancing the interaction. The final mood prediction is computed using emotion trends across all user responses, followed by curated music playlist recommendations. Initial Results: The implemented emotion classification model achieved an accuracy of 96.09%, with precision, recall, and F1 scores all around 0.96, indicating strong predictive performance and reliable emotion recognition.