Publication result detail

13 years of speaker recognition research at BUT, with longitudinal analysis of NIST SRE

MATĚJKA, P.; PLCHOT, O.; GLEMBEK, O.; BURGET, L.; ROHDIN, J.; ZEINALI, H.; MOŠNER, L.; SILNOVA, A.; NOVOTNÝ, O.; DIEZ SÁNCHEZ, M.; ČERNOCKÝ, J.

Original Title

13 years of speaker recognition research at BUT, with longitudinal analysis of NIST SRE

English Title

13 years of speaker recognition research at BUT, with longitudinal analysis of NIST SRE

Type

WoS Article

Original Abstract

In this paper, we present a brief history and a "longitudinal study" of all important milestonemodelling techniques used in text independent speaker recognition since Brno University ofTechnology (BUT) first participated in the NIST Speaker Recognition Evaluation (SRE) in2006-GMM MAP, GMM MAP with eigen-channel adaptation, Joint Factor Analysis, i-vectorand DNN embedding (x-vector). To emphasize the historical context, the techniques areevaluated on all NIST SRE sets since 2004 on a time-machine principle, i.e. a system is alwaystrained using all data available up till the year of evaluation. Moreover, as user-contributedaudiovisual content dominates nowadays Internet, we representatively include the SpeakersIn The Wild (SITW) and VOiCES challenge datasets in the evaluation of our systems. Not onlywe present a comparison of the modelling techniques, but we also show the effect of samplingfrequency.

English abstract

In this paper, we present a brief history and a "longitudinal study" of all important milestonemodelling techniques used in text independent speaker recognition since Brno University ofTechnology (BUT) first participated in the NIST Speaker Recognition Evaluation (SRE) in2006-GMM MAP, GMM MAP with eigen-channel adaptation, Joint Factor Analysis, i-vectorand DNN embedding (x-vector). To emphasize the historical context, the techniques areevaluated on all NIST SRE sets since 2004 on a time-machine principle, i.e. a system is alwaystrained using all data available up till the year of evaluation. Moreover, as user-contributedaudiovisual content dominates nowadays Internet, we representatively include the SpeakersIn The Wild (SITW) and VOiCES challenge datasets in the evaluation of our systems. Not onlywe present a comparison of the modelling techniques, but we also show the effect of samplingfrequency.

Keywords

Speaker recognition, NIST, Evaluations, GMM, Eigen-channel, compensation, JFA, I-vectors, DNN Embedding, X-vectors

Key words in English

Speaker recognition, NIST, Evaluations, GMM, Eigen-channel, compensation, JFA, I-vectors, DNN Embedding, X-vectors

Authors

MATĚJKA, P.; PLCHOT, O.; GLEMBEK, O.; BURGET, L.; ROHDIN, J.; ZEINALI, H.; MOŠNER, L.; SILNOVA, A.; NOVOTNÝ, O.; DIEZ SÁNCHEZ, M.; ČERNOCKÝ, J.

RIV year

2020

Released

01.09.2020

ISBN

0885-2308

Periodical

COMPUTER SPEECH AND LANGUAGE

Volume

2020

Number

63

State

United Kingdom of Great Britain and Northern Ireland

Pages from

1

Pages to

15

Pages count

15

URL

https://www.sciencedirect.com/science/article/pii/S0885230819302797?via%3Dihub

BibTex

@article{BUT162674,
  author="Pavel {Matějka} and Oldřich {Plchot} and Ondřej {Glembek} and Lukáš {Burget} and Johan Andréas {Rohdin} and Hossein {Zeinali} and Ladislav {Mošner} and Anna {Silnova} and Ondřej {Novotný} and Mireia {Diez Sánchez} and Jan {Černocký}",
  title="13 years of speaker recognition research at BUT, with longitudinal analysis of NIST SRE",
  journal="COMPUTER SPEECH AND LANGUAGE",
  year="2020",
  volume="2020",
  number="63",
  pages="1--15",
  doi="10.1016/j.csl.2019.101035",
  issn="0885-2308",
  url="https://www.sciencedirect.com/science/article/pii/S0885230819302797?via%3Dihub"
}

Documents

matejka_elsevier2019_101035