Musical Source Separation: An Introduction

Estefanía Cano, Derry Fitzgerald, Antoine Liutkus, Mark D. Plumbley, Fabian-Robert Stöter

In the general music source separation problem, we are given one or more signals that contain different mixtures of some original source signals. The task is to recover one or more of the source signals given the mixtures.

The figure below shows a spectrogram of an audio mix where the different musical sources are displayed with different colors. Mixture signals like this one are the starting point of MSS tasks.

 

How can we separate musical sources when their spectral content highly overlaps in the spectral domain?