Mood-Based Song Recommendation System

A study of existing music systems showed that many music applications fail to recommend songs based on emotion; instead, they use the user's history to recommend songs. The objective of this research work is to extract emotions from real-time input of the human face and suggest Bollywood songs based on the detected emotion and other factors. According to the results of the survey taken for this research work, users tend to listen to songs considering factors, genres, era, singer, time of the day, and activity. The requirements for the proposed system were the preparation of the dataset and finding relationships between different factors considered for the research work. The proposed system includes identification of emotion from the human face and song filtration based on the algorithm developed using the factors considered. Thus, the proposed system is an interactive mood-based songs recommendation system that considers the current emotion of the user along with vital factors related to songs and user preferences for these factors while recommending songs. In future, this song recommendation system can be improved by adding a bigger dataset of Bollywood and regional songs in major Indian languages.