CHATBOT EMOTION RECOGNITION AND MUSIC RECOMMENDATION

Abstract

In recent years, emotion-aware systems have gained significant importance in improving human-computer interaction. This project proposes a Chatbot Emotion Recognition and Music Recommendation System that detects user emotions through multiple input modes such as text, voice, and facial expressions using deep learning techniques. The system utilizes advanced models including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), and hybrid LSTM+GRU models combined with BERT embeddings for accurate emotion classification. The models are trained on a publicly available dataset from Kaggle Emotion Dataset, which contains labeled text data for different emotional states. The system evaluates model performance using metrics such as accuracy, precision, recall, and F1- score. Experimental results show that the BiLSTM model achieves the highest accuracy of 91.84%, outperforming other models. The system is implemented using Jupyter Notebook for model training and a web-based interface for real-time prediction. Based on the detected emotion, the chatbot recommends appropriate music to enhance user mood. The system supports multimodal interaction, allowing users to input text, voice, or facial expressions. This approach improves user experience and provides personalized recommendations. The proposed system demonstrates the effectiveness of deep learning in emotion recognition and its application in intelligent recommendation systems.