Development of a Music Recommender System to Promote Emotional Well-being

Antonio Ardura Carnicero. (2023). Development of a Music Recommender System to Promote Emotional Well-being. Trabajo Fin de Titulación (TFG). Universidad Politécnica de Madrid, ETSI Telecomunicación.

In recent years there has been a great advance in the study of human emotions and their regulation. Emotions can manifest themselves in different ways and affect different areas of people's lives, not only intrapersonal relationships, but also interpersonal ones. Regulating emotions is essential for well-being and daily functioning. When someone learns to control their emotions, they are better able to manage stress, maintain healthy relationships and make good decisions. Emotion regulation also helps prevent impulsive behaviors and fosters emotional stability and resilience, which are crucial to maintaining good mental health. Traditionally, techniques such as breath control or meditation have been used to achieve an adequate emotional management. However, new techniques based on artificial intelligence are currently being explored. One of the most popular solutions is the use of music to influence our emotional state. Existing solutions usually lack personalization and precision in the use of music for emotional regulation. Therefore, a project is proposed to fill these gaps by developing a music recommender system based on machine learning and evaluating the user's emotional state in real time. Additionally, this system addresses the problem of subjectivity in the interpretation of music for emotional regulation by incorporating machine learning techniques to recommend music based on each user's individual preferences and needs. In addition to the main objective of the project, which is the development of an emotional regulation system capable of recommending music tailored to the user's emotion, the project includes the following sub-objectives: - The identification of the system requirements, its design and the development of the application integrating the Spotify API. - The implementation of emotion analysis techniques to detect user emotion. - The use of machine learning techniques to learn and recommend appropriate music for each specific emotion. - The consideration of the user's personal preferences and musical characteristics when recommending music. The project will start with an analysis of the state of the art in relation to emotional regulation, emotion analysis and machine learning techniques. Then, the system requirements will be defined and the architecture will be designed. Tools such as Flutter and the Spotify API will be used to develop the application. Finally, tests and evaluations will be carried out to measure the effectiveness of the system in emotional regulation and the recommendation of music adapted to the user's emotion. To address this problem, various technologies, such as sentiment analysis and machine learning, will be used to develop a system that can detect the user's emotional state and provide them with a suitable playlist to help them regulate their emotions. Social networks will also be explored to better understand how music affects people's emotions and how it can be used more effectively.