Result Details

Evaluation of SpeakerBeam target speech extraction in real noisy and reverberant conditions

DELCROIX, M.; ŽMOLÍKOVÁ, K.; OCHIAI, T.; KINOSHITA, K.; ARAKI, S.; NAKATANI, T. Evaluation of SpeakerBeam target speech extraction in real noisy and reverberant conditions. THE JOURNAL OF THE ACOUSTICAL SOCIETY OF JAPAN, 2019, vol. 2019, no. 2, p. 1-2. ISSN: 0369-4232.
Type
journal article
Language
English
Authors
Delcroix Marc, FIT (FIT)
Žmolíková Kateřina, Ing., Ph.D., DCGM (FIT)
OCHIAI, T.
Kinoshita Keisuke, FIT (FIT)
ARAKI, S.
Nakatani Tomohiro, FIT (FIT)
Abstract

The paper discusses SpeakerBeam for target speaker extraction. The main variation is in the use of a scaling adaptation layer to adapt a neural network instead of the previously proposed factorized layer. This enables to greatly reduce the number of model parameters, without affecting performance.

Keywords

SpeakerBeam, speaker, extraction, neural network

URL
Published
2019
Pages
1–2
Journal
THE JOURNAL OF THE ACOUSTICAL SOCIETY OF JAPAN, vol. 2019, no. 2, ISSN 0369-4232
BibTeX 
@article{BUT160008,
  author="DELCROIX, M. and ŽMOLÍKOVÁ, K. and OCHIAI, T. and KINOSHITA, K. and ARAKI, S. and NAKATANI, T.",
  title="Evaluation of SpeakerBeam target speech extraction in real noisy and reverberant conditions",
  journal="THE JOURNAL OF THE ACOUSTICAL SOCIETY OF JAPAN",
  year="2019",
  volume="2019",
  number="2",
  pages="1--2",
  issn="0369-4232",
  url="https://www.fit.vut.cz/research/publication/12103/"
}
Files
Projects
IT4Innovations excellence in science, MŠMT, Národní program udržitelnosti II, LQ1602, start: 2016-01-01, end: 2020-12-31, completed
NTT - Speech enhancement front-end for robust automatic speech recognition with large amount of training data, NTT, start: 2019-01-01, end: 2019-12-31, completed
Research groups
Speech Data Mining Research Group BUT Speech@FIT (RG SPEECH)
Departments
Department of Computer Graphics and Multimedia (DCGM)
Back to top