Project Details

Project title: An investigation into the use of machine learning techniques for music recommendation based on emotional content: the case of automated labelling of songs based on emotional content

Course Title: Bsc Computer Science

Student’s Name: Eugene Afful

Supervisor’s Name: Emmett Cooper

Project type: Computer and Data Science

Project ID: 122

To design, develop and evaluate a system that will investigate how Machine Learning techniques can be used to appropriately identify the emotional context of music

The purpose of this project is to determine whether machine learning techniques can be used to gain the emotional context of music, therefore allowing for sad music to be labelled by a genre based on the emotion (mood). This can then be used to create a recommendation system for users. The project involved using lyrical sentiment, valence and arousal values to determine the overall mood of a song, which was then converted to a label. These were then used for song recommendations and refined to be more precise.

The benefits of this project include expanding on the new MER (Music Emotion Recognition) field by providing research and results into a music recommendation system based on MER. Its innovation comes from using machine learning techniques in the MER field to make a music recommendation system based solely on this, which can later be expanded upon.