University of Surrey

Test tubes in the lab Research in the ATI Dance Research

Referenceless Performance Evaluation of Audio Source Separation using Deep Neural Networks

Grais, Emad M., Wierstorf, Hagen, Ward, Dominic, Mason, Russell and Plumbley, Mark (2019) Referenceless Performance Evaluation of Audio Source Separation using Deep Neural Networks In: 27th European Signal Processing Conference (EUSIPCO), 2019-09-02-2019-09-06, A Coruña, Spain.

[img]
Preview
Text
GE_PID5988743.pdf - Accepted version Manuscript

Download (206kB) | Preview
Official URL: http://eusipco2019.org/

Abstract

Current performance evaluation for audio source separation depends on comparing the processed or separated signals with reference signals. Therefore, common performance evaluation toolkits are not applicable to real-world situations where the ground truth audio is unavailable. In this paper, we propose a performance evaluation technique that does not require reference signals in order to assess separation quality. The proposed technique uses a deep neural network (DNN) to map the processed audio into its quality score. Our experiment results show that the DNN is capable of predicting the sources-to-artifacts ratio from the blind source separation evaluation toolkit [1] for singing-voice separation without the need for reference signals.

Item Type: Conference or Workshop Item (Conference Paper)
Divisions : Faculty of Engineering and Physical Sciences > Electronic Engineering
Authors :
NameEmailORCID
Grais, Emad M.grais@surrey.ac.uk
Wierstorf, Hagen
Ward, Dominic
Mason, Russell
Plumbley, Markm.plumbley@surrey.ac.uk
Date : 3 June 2019
Funders : Engineering and Physical Sciences Research Council (EPSRC)
DOI : 10.23919/EUSIPCO.2019.8902932
Copyright Disclaimer : Copyright © 2019 by EURASIP. All rights reserved. IEEE is granted non-exclusive, irrevocable, royaltyfree worldwide rights to publish, sell and distribute this copyrighted work and any content derived from the copyrighted work in any format or media without restriction.
Uncontrolled Keywords : Performance evaluation; Deep learning; Audio source separation; BSS-Eval sources-to-artifacts ratio
Depositing User : Diane Maxfield
Date Deposited : 24 Jun 2019 11:03
Last Modified : 18 Feb 2020 15:58
URI: http://epubs.surrey.ac.uk/id/eprint/852063

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year


Information about this web site

© The University of Surrey, Guildford, Surrey, GU2 7XH, United Kingdom.
+44 (0)1483 300800