Audio based Music Classification based on Genre and Emotion using Gaussian Process

Abstract: In the field of Music Information Retrieval (MIR), music genre classification and music emotion recognition are the two main tasks to investigate for further development. In this project work, these two tasks are focused. For this purpose, Gaussian Processes Model is used. Gaussian Processes (GPs) are Bayesian nonparametric models that are becoming more popular for their superior capabilities to capture highly nonlinear data relationships in various tasks, such as dimensionality reduction, time series analysis, novelty detection, as well as classical regression and classification tasks. Gaussian Processes are used to investigate the feasibility and applicability of Gaussian Process model for music genre classification and emotion estimation. Along with this, we are reducing the time required for feature extraction for classification tasks. Principle component analysis (PCA) technique is used to reduce this time. In this, it selects and considers only higher order features for classification.

Keywords: Music Information Retrieval, Genres, Emotions, Features.