Detail publikace

Pairwise Discriminative Speaker Verification in the I -Vector Space

Originální název

Pairwise Discriminative Speaker Verification in the I -Vector Space

Anglický název

Pairwise Discriminative Speaker Verification in the I -Vector Space

Jazyk

en

Originální abstrakt

In this work we present a novel framework for discriminative training of speaker verification systems, where a trial is represented, as in the PLDA approach, by an i-vector pair, and the task is discrimination between same-speaker and dif- ferent-speaker classes. This pairwise SVM approach provides a more natural paradigm to speaker verification compared to the classical one-vs-all discriminative training.

Anglický abstrakt

In this work we present a novel framework for discriminative training of speaker verification systems, where a trial is represented, as in the PLDA approach, by an i-vector pair, and the task is discrimination between same-speaker and dif- ferent-speaker classes. This pairwise SVM approach provides a more natural paradigm to speaker verification compared to the classical one-vs-all discriminative training.

BibTex


@article{BUT103568,
  author="Sandro {Cumani} and Niko {Brummer} and Lukáš {Burget} and Pietro {Laface} and Oldřich {Plchot} and Vasileios {Vasilakakis}",
  title="Pairwise Discriminative Speaker Verification in the I -Vector Space",
  annote="In this work we present a novel framework for discriminative training of speaker
verification systems, where a trial is represented, as in the PLDA approach, by
an i-vector pair, and the task is discrimination between same-speaker and dif-
ferent-speaker classes. This pairwise SVM approach provides a more natural
paradigm to speaker verification compared to the classical one-vs-all
discriminative training.",
  address="NEUVEDEN",
  chapter="103568",
  doi="10.1109/TASL.2013.2245655",
  edition="NEUVEDEN",
  howpublished="print",
  institution="NEUVEDEN",
  number="6",
  volume="2013",
  year="2013",
  month="february",
  pages="1217--1227",
  publisher="NEUVEDEN",
  type="journal article - other"
}