Soft-Mask De-Mixing for Anechoic Mixtures

Swarnadeep Bagchi; Ruairí de Fréin

doi:10.1109/ISSC55427.2022.9826179

Soft-Mask De-Mixing for Anechoic Mixtures

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

This paper extends a computationally efficient, soft-mask based source separation (SS) technique called Redress, to anechoic mixing scenarios. SS methods are an integral part of hearing aid research. We call the resulting method D-Redress. In its original form, Redress was intended for instantaneous mixing scenarios. Numerical evaluations demonstrate that soft-mask based techniques reduce the level of artifacts in the separated speech. Monte Carlo trials on 1000 real speech mixtures demonstrate that the D-Redress successfully extends Redress in terms of Overall-Perceptual (OPS), Target-Perceptual (TPS) scores and Human-Ear Intelligibility (HEI).

Original language	English
Title of host publication	2022 33rd Irish Signals and Systems Conference, ISSC 2022
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781665452274
DOIs	https://doi.org/10.1109/ISSC55427.2022.9826179
Publication status	Published - 2022
Event	33rd Irish Signals and Systems Conference, ISSC 2022 - Cork, Ireland Duration: 9 Jun 2022 → 10 Jun 2022

Publication series

Name	2022 33rd Irish Signals and Systems Conference, ISSC 2022

Conference

Conference	33rd Irish Signals and Systems Conference, ISSC 2022
Country/Territory	Ireland
City	Cork
Period	9/06/22 → 10/06/22

Keywords

Human-Ear Intelligibility (HEI)
Overall-Perceptual Score (OPS)
Source-Separation
Target-Perceptual Score (TPS)
anechoic
relative attenuation

Access to Document

10.1109/ISSC55427.2022.9826179Licence: CC BY-NC-ND

https://arrow.tudublin.ie/creaart/139Licence: CC BY-NC-SA

Cite this

@inproceedings{657f8f4e1ce54e0cb7115deb9b9a2ec3,

title = "Soft-Mask De-Mixing for Anechoic Mixtures",

abstract = "This paper extends a computationally efficient, soft-mask based source separation (SS) technique called Redress, to anechoic mixing scenarios. SS methods are an integral part of hearing aid research. We call the resulting method D-Redress. In its original form, Redress was intended for instantaneous mixing scenarios. Numerical evaluations demonstrate that soft-mask based techniques reduce the level of artifacts in the separated speech. Monte Carlo trials on 1000 real speech mixtures demonstrate that the D-Redress successfully extends Redress in terms of Overall-Perceptual (OPS), Target-Perceptual (TPS) scores and Human-Ear Intelligibility (HEI).",

keywords = "Human-Ear Intelligibility (HEI), Overall-Perceptual Score (OPS), Source-Separation, Target-Perceptual Score (TPS), anechoic, relative attenuation",

author = "Swarnadeep Bagchi and {de Fr{\'e}in}, Ruair{\'i}",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 33rd Irish Signals and Systems Conference, ISSC 2022 ; Conference date: 09-06-2022 Through 10-06-2022",

year = "2022",

doi = "10.1109/ISSC55427.2022.9826179",

language = "English",

series = "2022 33rd Irish Signals and Systems Conference, ISSC 2022",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "2022 33rd Irish Signals and Systems Conference, ISSC 2022",

address = "United States",

}

Bagchi, S & de Fréin, R 2022, Soft-Mask De-Mixing for Anechoic Mixtures. in 2022 33rd Irish Signals and Systems Conference, ISSC 2022. 2022 33rd Irish Signals and Systems Conference, ISSC 2022, Institute of Electrical and Electronics Engineers Inc., 33rd Irish Signals and Systems Conference, ISSC 2022, Cork, Ireland, 9/06/22. https://doi.org/10.1109/ISSC55427.2022.9826179

Soft-Mask De-Mixing for Anechoic Mixtures. / Bagchi, Swarnadeep; de Fréin, Ruairí.
2022 33rd Irish Signals and Systems Conference, ISSC 2022. Institute of Electrical and Electronics Engineers Inc., 2022. (2022 33rd Irish Signals and Systems Conference, ISSC 2022).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Soft-Mask De-Mixing for Anechoic Mixtures

AU - Bagchi, Swarnadeep

AU - de Fréin, Ruairí

PY - 2022

Y1 - 2022

N2 - This paper extends a computationally efficient, soft-mask based source separation (SS) technique called Redress, to anechoic mixing scenarios. SS methods are an integral part of hearing aid research. We call the resulting method D-Redress. In its original form, Redress was intended for instantaneous mixing scenarios. Numerical evaluations demonstrate that soft-mask based techniques reduce the level of artifacts in the separated speech. Monte Carlo trials on 1000 real speech mixtures demonstrate that the D-Redress successfully extends Redress in terms of Overall-Perceptual (OPS), Target-Perceptual (TPS) scores and Human-Ear Intelligibility (HEI).

AB - This paper extends a computationally efficient, soft-mask based source separation (SS) technique called Redress, to anechoic mixing scenarios. SS methods are an integral part of hearing aid research. We call the resulting method D-Redress. In its original form, Redress was intended for instantaneous mixing scenarios. Numerical evaluations demonstrate that soft-mask based techniques reduce the level of artifacts in the separated speech. Monte Carlo trials on 1000 real speech mixtures demonstrate that the D-Redress successfully extends Redress in terms of Overall-Perceptual (OPS), Target-Perceptual (TPS) scores and Human-Ear Intelligibility (HEI).

KW - Human-Ear Intelligibility (HEI)

KW - Overall-Perceptual Score (OPS)

KW - Source-Separation

KW - Target-Perceptual Score (TPS)

KW - anechoic

KW - relative attenuation

UR - http://www.scopus.com/inward/record.url?scp=85135943004&partnerID=8YFLogxK

U2 - 10.1109/ISSC55427.2022.9826179

DO - 10.1109/ISSC55427.2022.9826179

M3 - Conference contribution

AN - SCOPUS:85135943004

T3 - 2022 33rd Irish Signals and Systems Conference, ISSC 2022

BT - 2022 33rd Irish Signals and Systems Conference, ISSC 2022

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 33rd Irish Signals and Systems Conference, ISSC 2022

Y2 - 9 June 2022 through 10 June 2022

ER -

Soft-Mask De-Mixing for Anechoic Mixtures

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this