Publication Details

Analysis of DNN-based Embeddings for Language Recognition on the NIST LRE 2017

LOZANO DÍEZ, A.; PLCHOT, O.; MATĚJKA, P.; NOVOTNÝ, O.; GONZALEZ-RODRIGUEZ, J. Analysis of DNN-based Embeddings for Language Recognition on the NIST LRE 2017. In Proceedings of Odyssey 2018 The Speaker and Language Recognition Workshop. Proceedings of Odyssey: The Speaker and Language Recognition Workshop Odyssey 2014, Joensuu, Finland. Les Sables d'Olonne: International Speech Communication Association, 2018. p. 39-46. ISSN: 2312-2846.
Czech title
Analýza DNN Embeddings pro rozpoznávání jazyka v NIST LRE 2017
Type
conference paper
Language
English
Authors
Lozano Díez Alicia, Ph.D.
Plchot Oldřich, Ing., Ph.D. (DCGM)
Matějka Pavel, Ing., Ph.D. (DCGM)
Novotný Ondřej, Ing., Ph.D.
Gonzalez-Rodriguez Joaquin
URL
Keywords

language recognition

Abstract

In this work, we analyze different designs of a language identification(LID) system based on embeddings. In our case, anembedding represents a whole utterance (or a speech segmentof variable duration) as a fixed-length vector (similar to the ivector).Moreover, this embedding aims to capture informationrelevant to the target task (LID), and it is obtained by training adeep neural network (DNN) to classify languages. In particular,we trained a DNN based on bidirectional long short-term memory(BLSTM) recurrent neural network (RNN) layers, whoseframe-by-frame outputs are summarized into mean and standarddeviation statistics for each utterance. After this pooling layer,we add two fully connected layers whose outputs are used asembeddings, which are afterwards modeled by a Gaussian linearclassifier (GLC). For training, we add a softmax output layerand train the whole network with multi-class cross-entropy objectiveto discriminate between languages. We analyze the effectof using data augmentation in the DNN training, as well asdifferent input features and architecture hyper-parameters, obtainingconfigurations that gradually improved the performanceof the embedding system. We report our results on the NISTLRE 2017 evaluation dataset and compare the performance ofembeddings with a reference i-vector system. We show thatthe best configuration of our embedding system outperforms thestrong reference i-vector system by 3% relative, and this is furtherpushed up to 10% relative improvement via a simple scorelevel fusion.

Published
2018
Pages
39–46
Journal
Proceedings of Odyssey: The Speaker and Language Recognition Workshop Odyssey 2014, Joensuu, Finland, vol. 2018, no. 6, ISSN 2312-2846
Proceedings
Proceedings of Odyssey 2018 The Speaker and Language Recognition Workshop
Conference
Odyssey 2018, Les Sables d'Olonne, France, FR
Publisher
International Speech Communication Association
Place
Les Sables d'Olonne
DOI
10.21437/Odyssey.2018-6
EID Scopus
2-s2.0-85083428874
BibTeX 
@inproceedings{BUT155066,
  author="Alicia {Lozano Díez} and Oldřich {Plchot} and Pavel {Matějka} and Ondřej {Novotný} and Joaquin {Gonzalez-Rodriguez}",
  title="Analysis of DNN-based Embeddings for Language Recognition on the NIST LRE 2017",
  booktitle="Proceedings of Odyssey 2018 The Speaker and Language Recognition Workshop",
  year="2018",
  journal="Proceedings of Odyssey: The Speaker and Language Recognition Workshop Odyssey 2014, Joensuu, Finland",
  volume="2018",
  number="6",
  pages="39--46",
  publisher="International Speech Communication Association",
  address="Les Sables d'Olonne",
  doi="10.21437/Odyssey.2018-6",
  issn="2312-2846",
  url="https://www.isca-speech.org/archive/Odyssey_2018/pdfs/42.pdf"
}
Files
Back to top