ai_system

2023-1 : Emotion-based Music Recommendation System

Goal

• Classify emotions from webcam input.
• Perform EDA and clustering on music dataset audio features.
• Recommend music based on vector similarity, aligning with the user's current emotional state and year-specific music preferences.

Dataset

• Emotion dataset : https://www.kaggle.com/datasets/msambare/fer2013
• Music dataset : https://www.kaggle.com/datasets/vatsalmavani/spotify-dataset

Algorithm

• Emotion Classification (Emotion_detection.ipynb)
• Audio Feature Clustering (EDA_Clustering.ipynb)

Run

python recommend.py

• Input: Emotion inference from webcam (5 seconds) & year selection.
• Output: Ranked list of top 1-5 song recommendations based on the input year and inferred emotion.
• Additional feature: Linking to YouTube for the recommended songs.