Detail publikace

BUT Text-Dependent Speaker Verification System for SdSV Challenge 2020

LOZANO DÍEZ, A. SILNOVA, A. PULUGUNDLA, B. ROHDIN, J. VESELÝ, K. BURGET, L. PLCHOT, O. GLEMBEK, O. NOVOTNÝ, O. MATĚJKA, P.

Originální název

BUT Text-Dependent Speaker Verification System for SdSV Challenge 2020

Anglický název

BUT Text-Dependent Speaker Verification System for SdSV Challenge 2020

Jazyk

en

Originální abstrakt

In this paper, we present the winning BUT submission for the text-dependent task of the SdSV challenge 2020. Given the large amount of training data available in this challenge, we explore successful techniques from text-independent systems in the text-dependent scenario. In particular, we trained x-vector extractors on both in-domain and out-of-domain datasets and combine them with i-vectors trained on concatenated MFCCs and bottleneck features, which have proven effective for the text-dependent scenario. Moreover, we proposed the use of phrase-dependent PLDA backend for scoring and its combination with a simple phrase recognizer, which brings up to 63% relative improvement on our development set with respect to using standard PLDA. Finally, we combine our different i-vector and x-vector based systems using a simple linear logistic regression score level fusion, which provides 28% relative improvement on the evaluation set with respect to our best single system.

Anglický abstrakt

In this paper, we present the winning BUT submission for the text-dependent task of the SdSV challenge 2020. Given the large amount of training data available in this challenge, we explore successful techniques from text-independent systems in the text-dependent scenario. In particular, we trained x-vector extractors on both in-domain and out-of-domain datasets and combine them with i-vectors trained on concatenated MFCCs and bottleneck features, which have proven effective for the text-dependent scenario. Moreover, we proposed the use of phrase-dependent PLDA backend for scoring and its combination with a simple phrase recognizer, which brings up to 63% relative improvement on our development set with respect to using standard PLDA. Finally, we combine our different i-vector and x-vector based systems using a simple linear logistic regression score level fusion, which provides 28% relative improvement on the evaluation set with respect to our best single system.

Dokumenty

BibTex


@inproceedings{BUT168145,
  author="Alicia {Lozano Díez} and Anna {Silnova} and Bhargav {Pulugundla} and Johan Andréas {Rohdin} and Karel {Veselý} and Lukáš {Burget} and Oldřich {Plchot} and Ondřej {Glembek} and Ondřej {Novotný} and Pavel {Matějka}",
  title="BUT Text-Dependent Speaker Verification System for SdSV Challenge 2020",
  annote="In this paper, we present the winning BUT submission for the text-dependent task
of the SdSV challenge 2020. Given the large amount of training data available in
this challenge, we explore successful techniques from text-independent systems in
the text-dependent scenario. In particular, we trained x-vector extractors on
both in-domain and out-of-domain datasets and combine them with i-vectors trained
on concatenated MFCCs and bottleneck features, which have proven effective for
the text-dependent scenario. Moreover, we proposed the use of phrase-dependent
PLDA backend for scoring and its combination with a simple phrase recognizer,
which brings up to 63% relative improvement on our development set with respect
to using standard PLDA. Finally, we combine our different i-vector and x-vector
based systems using a simple linear logistic regression score level fusion, which
provides 28% relative improvement on the evaluation set with respect to our best
single system.",
  address="International Speech Communication Association",
  booktitle="Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH",
  chapter="168145",
  doi="10.21437/Interspeech.2020-2882",
  edition="NEUVEDEN",
  howpublished="online",
  institution="International Speech Communication Association",
  number="10",
  year="2020",
  month="october",
  pages="761--765",
  publisher="International Speech Communication Association",
  type="conference paper"
}