Emotion-Driven Music Recommender System with Deep Learning and Streamlit Integration

The emotion-driven music recommendation system suggests music by looking at the detected emotions. It uses a webcam to check the facial emotions. Based on the emotion detected, the system picks songs from YouTube or Spotify. This system has been trained on the dataset Facial Emotion Recognition (FER-2013) to detect emotions. The emotions detected are early and precise. Recommended approach mainly uses two models they are Convolutional Neural Networks (CNN), and Multi-Task Convolutional Neural Networks (MTCNN's). Designed method mainly focuses on identifying the user's current emotion state, instead of only relying on the past music choices. By connecting with YouTube and Spotify, it can offer a lot of different music, making the suggestions more personal. The system becomes more attentive to people's various emotions. Several emotions such as anger, disgust, fear, happy, neutral, sad, and surprise are trained using this model. Introduced system achieved a validation accuracy of 96% and a testing accuracy of 94% in detecting emotions. It effectively recommends mood-aligned songs, ensuring a responsive, real-time experience. The proposed model demonstrates the successful integration of deep learning and real-time music recommendation, offering users an intuitive and personalized experience based on their emotional state. The combination of CNN, MTCNN and streaming APIs provides a scalable and efficient solution.