Publication detail

Analysis of the DNN-Based SRE Systems in Multi-language Conditions

NOVOTNÝ, O. MATĚJKA, P. GLEMBEK, O. PLCHOT, O. GRÉZL, F. BURGET, L. ČERNOCKÝ, J.

Original Title

Analysis of the DNN-Based SRE Systems in Multi-language Conditions

Type

conference paper

Language

English

Original Abstract

This paper analyzes the behavior of our state-of-the-art Deep Neural Network/i-vector/PLDA-based speaker recognition systems in multi-language conditions. On the "Language Pack" of the PRISM set, we evaluate the systems performance using the NISTs standard metrics. We show that not only the gain from using DNNs vanishes, nor using dedicated DNNs for target conditions helps, but also the DNN-based systems tend to produce de-calibrated scores under the studied conditions. This work gives suggestions for directions of future research rather than any particular solutions to these issues.

Keywords

DNN, Multi-Language, Speaker Recognition

Authors

NOVOTNÝ, O.; MATĚJKA, P.; GLEMBEK, O.; PLCHOT, O.; GRÉZL, F.; BURGET, L.; ČERNOCKÝ, J.

Released

13. 12. 2016

Publisher

IEEE Signal Processing Society

Location

San Diego

ISBN

978-1-5090-4903-5

Book

Proceedings of SLT 2016

Pages from

199

Pages to

204

Pages count

6

URL

http://ieeexplore.ieee.org/document/7846265/

BibTex

@inproceedings{BUT132603,
  author="Ondřej {Novotný} and Pavel {Matějka} and Ondřej {Glembek} and Oldřich {Plchot} and František {Grézl} and Lukáš {Burget} and Jan {Černocký}",
  title="Analysis of the DNN-Based SRE Systems in Multi-language Conditions",
  booktitle="Proceedings of SLT 2016",
  year="2016",
  pages="199--204",
  publisher="IEEE Signal Processing Society",
  address="San Diego",
  doi="10.1109/slt.2016.7846265",
  isbn="978-1-5090-4903-5",
  url="http://ieeexplore.ieee.org/document/7846265/"
}