Psychophysical Evaluation of Audio Source Separation Methods

Andrew JR Simpson, Gerard Roma, Emad M Grais, Russell D Mason, Christopher Hummersone, Mark D Plumbley

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

Source separation evaluation is typically a top-down process, starting with perceptual measures which capture fitness-for-purpose and followed by attempts to find physical (objective) measures that are predictive of the perceptual measures. In this paper, we take a contrasting bottom-up approach. We begin with the physical measures provided by the Blind Source Separation Evaluation Toolkit (BSS Eval) and we then look for corresponding perceptual correlates. This approach is known as psychophysics and has the distinct advantage of leading to interpretable, psychophysical models. We obtained perceptual similarity judgments from listeners in two experiments featuring vocal sources within musical mixtures. In the first experiment, listeners compared the overall quality of vocal signals estimated from musical mixtures using a range of competing source separation methods. In a loudness experiment, listeners compared the loudness balance of the competing musical accompaniment and vocal. Our preliminary results provide provisional validation of the psychophysical approach.
Original languageEnglish
Title of host publicationLatent Variable Analysis and Signal Separation
EditorsPetr Tichavský, Massoud Babaie-Zadeh, Olivier J.J. Michel, Nadège Thirion-Moreau
PublisherSpringer, Cham
Pages211-221
Number of pages11
ISBN (Electronic)9783319535470
ISBN (Print)9783319535463
DOIs
Publication statusPublished - 15 Feb 2017
Externally publishedYes
Event13th International Conference on Latent Variable Analysis and Signal Separation - Grenoble-Alpes University, Grenoble, France
Duration: 21 Feb 201723 Feb 2017
Conference number: 13
http://www.lva-ica-2017.com/ (Link to Conference Website)

Publication series

Name Lecture Notes in Computer Science
PublisherSpringer
Volume10169
ISSN (Electronic)0302-9743

Conference

Conference13th International Conference on Latent Variable Analysis and Signal Separation
Abbreviated titleLVA/ICA 2017
CountryFrance
CityGrenoble
Period21/02/1723/02/17
Internet address

Fingerprint

Source separation
Blind source separation
Experiments

Cite this

Simpson, A. JR., Roma, G., Grais, E. M., Mason, R. D., Hummersone, C., & Plumbley, M. D. (2017). Psychophysical Evaluation of Audio Source Separation Methods. In P. Tichavský, M. Babaie-Zadeh, O. J. J. Michel, & N. Thirion-Moreau (Eds.), Latent Variable Analysis and Signal Separation (pp. 211-221). ( Lecture Notes in Computer Science; Vol. 10169). Springer, Cham. https://doi.org/10.1007/978-3-319-53547-0_21
Simpson, Andrew JR ; Roma, Gerard ; Grais, Emad M ; Mason, Russell D ; Hummersone, Christopher ; Plumbley, Mark D. / Psychophysical Evaluation of Audio Source Separation Methods. Latent Variable Analysis and Signal Separation. editor / Petr Tichavský ; Massoud Babaie-Zadeh ; Olivier J.J. Michel ; Nadège Thirion-Moreau. Springer, Cham, 2017. pp. 211-221 ( Lecture Notes in Computer Science).
@inproceedings{b67bb62313b442d980af37c866871a2e,
title = "Psychophysical Evaluation of Audio Source Separation Methods",
abstract = "Source separation evaluation is typically a top-down process, starting with perceptual measures which capture fitness-for-purpose and followed by attempts to find physical (objective) measures that are predictive of the perceptual measures. In this paper, we take a contrasting bottom-up approach. We begin with the physical measures provided by the Blind Source Separation Evaluation Toolkit (BSS Eval) and we then look for corresponding perceptual correlates. This approach is known as psychophysics and has the distinct advantage of leading to interpretable, psychophysical models. We obtained perceptual similarity judgments from listeners in two experiments featuring vocal sources within musical mixtures. In the first experiment, listeners compared the overall quality of vocal signals estimated from musical mixtures using a range of competing source separation methods. In a loudness experiment, listeners compared the loudness balance of the competing musical accompaniment and vocal. Our preliminary results provide provisional validation of the psychophysical approach.",
keywords = "Deep learning, Perceptual evaluation, Source separation",
author = "Simpson, {Andrew JR} and Gerard Roma and Grais, {Emad M} and Mason, {Russell D} and Christopher Hummersone and Plumbley, {Mark D}",
year = "2017",
month = "2",
day = "15",
doi = "10.1007/978-3-319-53547-0_21",
language = "English",
isbn = "9783319535463",
series = "Lecture Notes in Computer Science",
publisher = "Springer, Cham",
pages = "211--221",
editor = "Petr Tichavsk{\'y} and Massoud Babaie-Zadeh and Michel, {Olivier J.J.} and Nad{\`e}ge Thirion-Moreau",
booktitle = "Latent Variable Analysis and Signal Separation",

}

Simpson, AJR, Roma, G, Grais, EM, Mason, RD, Hummersone, C & Plumbley, MD 2017, Psychophysical Evaluation of Audio Source Separation Methods. in P Tichavský, M Babaie-Zadeh, OJJ Michel & N Thirion-Moreau (eds), Latent Variable Analysis and Signal Separation. Lecture Notes in Computer Science, vol. 10169, Springer, Cham, pp. 211-221, 13th International Conference on Latent Variable Analysis and Signal Separation, Grenoble, France, 21/02/17. https://doi.org/10.1007/978-3-319-53547-0_21

Psychophysical Evaluation of Audio Source Separation Methods. / Simpson, Andrew JR; Roma, Gerard; Grais, Emad M; Mason, Russell D; Hummersone, Christopher; Plumbley, Mark D.

Latent Variable Analysis and Signal Separation. ed. / Petr Tichavský; Massoud Babaie-Zadeh; Olivier J.J. Michel; Nadège Thirion-Moreau. Springer, Cham, 2017. p. 211-221 ( Lecture Notes in Computer Science; Vol. 10169).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Psychophysical Evaluation of Audio Source Separation Methods

AU - Simpson, Andrew JR

AU - Roma, Gerard

AU - Grais, Emad M

AU - Mason, Russell D

AU - Hummersone, Christopher

AU - Plumbley, Mark D

PY - 2017/2/15

Y1 - 2017/2/15

N2 - Source separation evaluation is typically a top-down process, starting with perceptual measures which capture fitness-for-purpose and followed by attempts to find physical (objective) measures that are predictive of the perceptual measures. In this paper, we take a contrasting bottom-up approach. We begin with the physical measures provided by the Blind Source Separation Evaluation Toolkit (BSS Eval) and we then look for corresponding perceptual correlates. This approach is known as psychophysics and has the distinct advantage of leading to interpretable, psychophysical models. We obtained perceptual similarity judgments from listeners in two experiments featuring vocal sources within musical mixtures. In the first experiment, listeners compared the overall quality of vocal signals estimated from musical mixtures using a range of competing source separation methods. In a loudness experiment, listeners compared the loudness balance of the competing musical accompaniment and vocal. Our preliminary results provide provisional validation of the psychophysical approach.

AB - Source separation evaluation is typically a top-down process, starting with perceptual measures which capture fitness-for-purpose and followed by attempts to find physical (objective) measures that are predictive of the perceptual measures. In this paper, we take a contrasting bottom-up approach. We begin with the physical measures provided by the Blind Source Separation Evaluation Toolkit (BSS Eval) and we then look for corresponding perceptual correlates. This approach is known as psychophysics and has the distinct advantage of leading to interpretable, psychophysical models. We obtained perceptual similarity judgments from listeners in two experiments featuring vocal sources within musical mixtures. In the first experiment, listeners compared the overall quality of vocal signals estimated from musical mixtures using a range of competing source separation methods. In a loudness experiment, listeners compared the loudness balance of the competing musical accompaniment and vocal. Our preliminary results provide provisional validation of the psychophysical approach.

KW - Deep learning

KW - Perceptual evaluation

KW - Source separation

U2 - 10.1007/978-3-319-53547-0_21

DO - 10.1007/978-3-319-53547-0_21

M3 - Conference contribution

SN - 9783319535463

T3 - Lecture Notes in Computer Science

SP - 211

EP - 221

BT - Latent Variable Analysis and Signal Separation

A2 - Tichavský, Petr

A2 - Babaie-Zadeh, Massoud

A2 - Michel, Olivier J.J.

A2 - Thirion-Moreau, Nadège

PB - Springer, Cham

ER -

Simpson AJR, Roma G, Grais EM, Mason RD, Hummersone C, Plumbley MD. Psychophysical Evaluation of Audio Source Separation Methods. In Tichavský P, Babaie-Zadeh M, Michel OJJ, Thirion-Moreau N, editors, Latent Variable Analysis and Signal Separation. Springer, Cham. 2017. p. 211-221. ( Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-319-53547-0_21