Evaluation of audio source separation models using hypothesis-driven non-parametric statistical methods

Andrew JR Simpson, Gerard Roma, Emad M Grais, Russell D Mason, Chris Hummersone, Antoine Liutkus, Mark D Plumbley

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

8 Citations (Scopus)


Audio source separation models are typically evaluated using objective separation quality measures, but rigorous statistical methods have yet to be applied to the problem of model comparison. As a result, it can be difficult to establish whether or not reliable progress is being made during the development of new models. In this paper, we provide a hypothesis-driven statistical analysis of the results of the recent source separation SiSEC challenge involving twelve competing models tested on separation of voice and accompaniment from fifty pieces of “professionally produced” contemporary music. Using non-parametric statistics, we establish reliable evidence for meaningful conclusions about the performance of the various models.
Original languageEnglish
Title of host publication2016 24th European Signal Processing Conference (EUSIPCO)
Number of pages5
ISBN (Electronic)9780992862657
Publication statusPublished - 1 Dec 2016
Externally publishedYes
Event24th European Signal Processing Conference - Budapest, Hungary
Duration: 29 Aug 20162 Sep 2016
Conference number: 24
http://www.eusipco2016.org/ (Link to Conference Website )

Publication series

ISSN (Electronic)2076-1465


Conference24th European Signal Processing Conference
Abbreviated titleEUSIPCO 2016
Internet address


Dive into the research topics of 'Evaluation of audio source separation models using hypothesis-driven non-parametric statistical methods'. Together they form a unique fingerprint.

Cite this