Detail publikace

Speaker Verification with Application-Aware Beamforming

MOŠNER, L. PLCHOT, O. ROHDIN, J. BURGET, L. ČERNOCKÝ, J.

Originální název

Speaker Verification with Application-Aware Beamforming

Typ

článek ve sborníku ve WoS nebo Scopus

Jazyk

angličtina

Originální abstrakt

Multichannel speech processing applications usually employbeamformers as means of speech enhancement through spatialfiltering. Beamformers with learnable parameters requiretraining to minimize a loss function that is not necessarilycorrelated with the final objective. In this paper, we presenta framework employing recent neural network based generalizedeigenvalue beamformer and application-specific modelthat allows for optimization of beamformer w.r.t. target application.In our case, the application is speaker verificationwhich utilizes a speaker embedding (x-vector) extractorthat conveniently comes with desired loss. We show thatapplication-specific training of the beamformer brings performanceimprovements over a system trained in the standardway. We perform our analysis on the recently introducedVOiCES corpus which contains multichannel data and allowsus to modify the evaluation trials such that enrollment recordingsremain single-channel and test utterances are multichannel.

Klíčová slova

Speaker verification, beamforming, xvector, generalized eigenvalue problem

Autoři

MOŠNER, L.; PLCHOT, O.; ROHDIN, J.; BURGET, L.; ČERNOCKÝ, J.

Vydáno

14. 12. 2019

Nakladatel

IEEE Signal Processing Society

Místo

Sentosa, Singapore

ISBN

978-1-7281-0306-8

Kniha

IEEE Automatic Speech Recognition and Understanding Workshop - Proceedings (ASRU)

Strany od

411

Strany do

418

Strany počet

8

URL

https://www.fit.vut.cz/research/publication/12152/

BibTex

@inproceedings{BUT161476,
  author="Ladislav {Mošner} and Oldřich {Plchot} and Johan Andréas {Rohdin} and Lukáš {Burget} and Jan {Černocký}",
  title="Speaker Verification with Application-Aware Beamforming",
  booktitle="IEEE Automatic Speech Recognition and Understanding Workshop - Proceedings (ASRU)",
  year="2019",
  pages="411--418",
  publisher="IEEE Signal Processing Society",
  address="Sentosa, Singapore",
  doi="10.1109/ASRU46091.2019.9003932",
  isbn="978-1-7281-0306-8",
  url="https://www.fit.vut.cz/research/publication/12152/"
}

Dokumenty

mosner_asru2019_0000411.pdf