To assist with the development of intelligent mixing systems, it would be useful to be able to extract the loudness balance of sources in an existing musical mixture. The relative-to-mix loudness level of four instrument groups was predicted using the sources extracted by 12 audio source separation algorithms. The predictions were compared with the ground truth loudness data of the original unmixed stems obtained from a recent dataset involving 100 mixed songs. It was found that the best source separation system could predict the relative loudness of each instrument group with an average root-mean-square error of 1.2 LU, with superior performance obtained on vocals.


  title = {Estimating the Loudness Balance of Musical Mixtures using Audio Source Separation},
  author = {Ward, D. and Wierstorf, H. and Mason, R. D. and Plumbley, M. D. and Hummersone, C.},
  booktitle = {3rd Workshop on Intelligent Music Production},
  address = {Salford, UK},
  month = sep,
  year = {2017},
  openaccess = {http://epubs.surrey.ac.uk/841966/},
  keywords = {"maruss"}

Supplementary material

All source code associated with this work are published at 10.5281/zenodo.1146318. This record also includes the presentation given in session A of WIMP 2017.

You can hg clone the project repository using the following URL: