Music Recommendation Based on Facial Expression
Mrudula K, Harsh R Jain, Amogha R Chandra, Jayanth Bhansali
B.E, Computer Science & Engineering, B.N.M Institute of Technology, Bangalore
Abstract – This paper describes various methods for music recommendation based on facial expressions. Multiple music applications suggest music based on a user’s music history. But there has been some research going on for music recommendation using facial expressions/mood detection. Facial expression recognition requires image processing, which can be done using multiple algorithms, along with feature extraction, and classified into different emotions. Music genre classification requires audio processing and genres are classified based on certain audio features, and some classification algorithms. This survey paper describes and compares multiple algorithms, for the same.
Keywords – Music recommendation, Facial expressions, Mood detection, Feature extraction, Image processing, Audio processing.
I. INTRODUCTION
Music is an experience beyond words. It is a part of every known society, past and present, and is common to all human cultures across the globe. It expresses a person’s feelings and mood. Everyone has their playlists defined for every mood, and certain media applications to use, like YouTube, Spotify, etc., and have witnessed song recommendations on these apps based on music history. Also, another form of non-verbal communication is facial expressions. According to a study, over 65% of human communication is non-verbal, which may include expressions, gestures, etc. Music not only heals a person’s mood but also has a great impact on one’s mood. Every person resorts to music, in almost every situation, and automatic detection of the user’s mood is an advancement to the currently available technologies. Music recommendation systems and Facial expression detection require the use of certain Deep Learning and Machine Learning models.
Machine learning, in simple words, is training a machine to learn and improve in performance based on prior information given to it. It is a branch of artificial intelligence (AI), that makes the machine capable of learning and becoming more accurate, without the user having to explicitly program it. Machinelearning algorithms read/collect data and learn at every step, for themselves.
Deep learning, on the other hand, is the branch of AI that replicates the functioning of the human brain to detect objects, speech recognition, language translation, etc.Deep learning algorithms also learn automatically from both labeled and unstructured data. Deep learning is used across