Paper Title
Matrices in Physics Data Processing

Abstract
Source separation plays an important role in audio processing. Hence, we would like to incorporate matrix factorisation feature extraction properties which are well-known in image processing into audio source separation. Therefore, a comparison between Nonnegative Matrix Factorisation, Convolutive Nonnegative Matrix Factorisation and Interpolative Decomposition in performing source separation and evaluate the performance using the Signal-to-Noise ratio. In the end, we found out that Convolutive Nonnegative Matrix Factorisation with Itakura-Saito divergence has better performance in separating the musical instruments as compared to other methods. Keywords - Audio Processing, Nonnegative Matrix Factorisation, Convolutive Nonnegative Matrix Factorisation, Interpolative Decomposition