A DUET-Based Method for Blind Separation of Speech Signals in Reverberant Environments

Minook KIM; Tae-Jun LEE; Hyung-Min PARK

doi:10.1587/transfun.E98.A.2325

A DUET-Based Method for Blind Separation of Speech Signals in Reverberant Environments

Minook KIM, Tae-Jun LEE, Hyung-Min PARK

Full Text Views

0

Share
Cite this

Summary :

This letter presents a two-stage method to extend the degenerate unmixing estimation technique (DUET) for reverberant speech separation. First, frequency-bin-wise attenuation and delay parameters are introduced and estimated by online update rules, to handle early reflections. Next, a mask reestimation algorithm based on the precedence effect is developed to detect and fix the errors on binary masks caused by late reflections. Experimental results demonstrate that the proposed method improves separation performance significantly.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E98-A No.11 pp.2325-2329

Publication Date: 2015/11/01

Publicized

Online ISSN: 1745-1337

DOI: 10.1587/transfun.E98.A.2325

Type of Manuscript: LETTER

Category: Speech and Hearing

Authors

Minook KIM
  Sogang University
Tae-Jun LEE
  Sogang University
Hyung-Min PARK
  Sogang University

Keyword

degenerate unmixing estimation technique, blind source separation, reverberation, time-frequency masking

Cite this

Copy

Minook KIM, Tae-Jun LEE, Hyung-Min PARK, "A DUET-Based Method for Blind Separation of Speech Signals in Reverberant Environments" in IEICE TRANSACTIONS on Fundamentals, vol. E98-A, no. 11, pp. 2325-2329, November 2015, doi: 10.1587/transfun.E98.A.2325.
Abstract: This letter presents a two-stage method to extend the degenerate unmixing estimation technique (DUET) for reverberant speech separation. First, frequency-bin-wise attenuation and delay parameters are introduced and estimated by online update rules, to handle early reflections. Next, a mask reestimation algorithm based on the precedence effect is developed to detect and fix the errors on binary masks caused by late reflections. Experimental results demonstrate that the proposed method improves separation performance significantly.
URL: https://globals.ieice.org/en_transactions/fundamentals/10.1587/transfun.E98.A.2325/_p

Copy

@ARTICLE{e98-a_11_2325,
author={Minook KIM, Tae-Jun LEE, Hyung-Min PARK, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={A DUET-Based Method for Blind Separation of Speech Signals in Reverberant Environments},
year={2015},
volume={E98-A},
number={11},
pages={2325-2329},
abstract={This letter presents a two-stage method to extend the degenerate unmixing estimation technique (DUET) for reverberant speech separation. First, frequency-bin-wise attenuation and delay parameters are introduced and estimated by online update rules, to handle early reflections. Next, a mask reestimation algorithm based on the precedence effect is developed to detect and fix the errors on binary masks caused by late reflections. Experimental results demonstrate that the proposed method improves separation performance significantly.},
keywords={},
doi={10.1587/transfun.E98.A.2325},
ISSN={1745-1337},
month={November},}

Copy

TY - JOUR
TI - A DUET-Based Method for Blind Separation of Speech Signals in Reverberant Environments
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 2325
EP - 2329
AU - Minook KIM
AU - Tae-Jun LEE
AU - Hyung-Min PARK
PY - 2015
DO - 10.1587/transfun.E98.A.2325
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E98-A
IS - 11
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - November 2015
AB - This letter presents a two-stage method to extend the degenerate unmixing estimation technique (DUET) for reverberant speech separation. First, frequency-bin-wise attenuation and delay parameters are introduced and estimated by online update rules, to handle early reflections. Next, a mask reestimation algorithm based on the precedence effect is developed to detect and fix the errors on binary masks caused by late reflections. Experimental results demonstrate that the proposed method improves separation performance significantly.
ER -