Repository logo
  • English
  • Català
  • Čeština
  • Deutsch
  • Español
  • Français
  • Gàidhlig
  • Italiano
  • Latviešu
  • Magyar
  • Nederlands
  • Polski
  • Português
  • Português do Brasil
  • Suomi
  • Svenska
  • Türkçe
  • Қазақ
  • বাংলা
  • हिंदी
  • Ελληνικά
  • Yкраї́нська
  • Log In
    or
    New user? Click here to register.Have you forgotten your password?
Repository logo
  • Communities & Collections
  • Research Outputs
  • Fundings & Projects
  • People
  • Statistics
  • English
  • Català
  • Čeština
  • Deutsch
  • Español
  • Français
  • Gàidhlig
  • Italiano
  • Latviešu
  • Magyar
  • Nederlands
  • Polski
  • Português
  • Português do Brasil
  • Suomi
  • Svenska
  • Türkçe
  • Қазақ
  • বাংলা
  • हिंदी
  • Ελληνικά
  • Yкраї́нська
  • Log In
    or
    New user? Click here to register.Have you forgotten your password?
  1. Home
  2. Indian Institute of Technology Madras
  3. Publication3
  4. ECAPA-TDNN embeddings for speaker diarization
 
  • Details
Options

ECAPA-TDNN embeddings for speaker diarization

Date Issued
01-01-2021
Author(s)
Dawalatabad, Nauman
Ravanelli, Mirco
Grondin, François
Thienpondt, Jenthe
Desplanques, Brecht
Na, Hwidong
DOI
10.21437/Interspeech.2021-941
Abstract
Learning robust speaker embeddings is a crucial step in speaker diarization. Deep neural networks can accurately capture speaker discriminative characteristics and popular deep embeddings such as x-vectors are nowadays a fundamental component of modern diarization systems. Recently, some improvements over the standard TDNN architecture used for x-vectors have been proposed. The ECAPA-TDNN model, for instance, has shown impressive performance in the speaker verification domain, thanks to a carefully designed neural model. In this work, we extend, for the first time, the use of the ECAPA-TDNN model to speaker diarization. Moreover, we improved its robustness with a powerful augmentation scheme that concatenates several contaminated versions of the same signal within the same training batch. The ECAPA-TDNN model turned out to provide robust speaker embeddings under both close-talking and distant-talking conditions. Our results on the popular AMI meeting corpus show that our system significantly outperforms recently proposed approaches.
Volume
4
Subjects
  • Data augmentation

  • Speaker diarization

  • Speaker embedding

  • Spectral clustering

Indian Institute of Technology Madras Knowledge Repository developed and maintained by the Library

Built with DSpace-CRIS software - Extension maintained and optimized by 4Science

  • Cookie settings
  • Privacy policy
  • End User Agreement
  • Send Feedback