Detail publikace

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.

Originální název

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

Anglický název

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

Jazyk

en

Originální abstrakt

In this work, we analyze different designs of a language identification (LID) system based on embeddings. In our case, an embedding represents a whole utterance (or a speech segment of variable duration) as a fixed-length vector (similar to the ivector). Moreover, this embedding aims to capture information relevant to the target task (LID), and it is obtained by training a deep 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, whose frame-by-frame outputs are summarized into mean and standard deviation statistics for each utterance. After this pooling layer, we add two fully connected layers whose outputs are used as embeddings, which are afterwards modeled by a Gaussian linear classifier (GLC). For training, we add a softmax output layer and train the whole network with multi-class cross-entropy objective to discriminate between languages. We analyze the effect of using data augmentation in the DNN training, as well as different input features and architecture hyper-parameters, obtaining configurations that gradually improved the performance of the embedding system. We report our results on the NIST LRE 2017 evaluation dataset and compare the performance of embeddings with a reference i-vector system. We show that the best configuration of our embedding system outperforms the strong reference i-vector system by 3% relative, and this is further pushed up to 10% relative improvement via a simple score level fusion.

Anglický abstrakt

In this work, we analyze different designs of a language identification (LID) system based on embeddings. In our case, an embedding represents a whole utterance (or a speech segment of variable duration) as a fixed-length vector (similar to the ivector). Moreover, this embedding aims to capture information relevant to the target task (LID), and it is obtained by training a deep 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, whose frame-by-frame outputs are summarized into mean and standard deviation statistics for each utterance. After this pooling layer, we add two fully connected layers whose outputs are used as embeddings, which are afterwards modeled by a Gaussian linear classifier (GLC). For training, we add a softmax output layer and train the whole network with multi-class cross-entropy objective to discriminate between languages. We analyze the effect of using data augmentation in the DNN training, as well as different input features and architecture hyper-parameters, obtaining configurations that gradually improved the performance of the embedding system. We report our results on the NIST LRE 2017 evaluation dataset and compare the performance of embeddings with a reference i-vector system. We show that the best configuration of our embedding system outperforms the strong reference i-vector system by 3% relative, and this is further pushed up to 10% relative improvement via a simple score level fusion.

Dokumenty

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",
  annote="In this work, we analyze different designs of a language identification (LID)
system based on embeddings. In our case, an embedding represents a whole
utterance (or a speech segment of variable duration) as a fixed-length vector
(similar to the ivector). Moreover, this embedding aims to capture information
relevant to the target task (LID), and it is obtained by training a deep 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, whose frame-by-frame outputs are summarized into mean and standard
deviation statistics for each utterance. After this pooling layer, we add two
fully connected layers whose outputs are used as embeddings, which are afterwards
modeled by a Gaussian linear classifier (GLC). For training, we add a softmax
output layer and train the whole network with multi-class cross-entropy objective
to discriminate between languages. We analyze the effect of using data
augmentation in the DNN training, as well as different input features and
architecture hyper-parameters, obtaining configurations that gradually improved
the performance of the embedding system. We report our results on the NIST LRE
2017 evaluation dataset and compare the performance of embeddings with
a reference i-vector system. We show that the best configuration of our embedding
system outperforms the strong reference i-vector system by 3% relative, and this
is further pushed up to 10% relative improvement via a simple score level
fusion.",
  address="International Speech Communication Association",
  booktitle="Proceedings of Odyssey 2018 The Speaker and Language Recognition Workshop",
  chapter="155066",
  doi="10.21437/Odyssey.2018-6",
  edition="NEUVEDEN",
  howpublished="online",
  institution="International Speech Communication Association",
  number="6",
  year="2018",
  month="june",
  pages="39--46",
  publisher="International Speech Communication Association",
  type="conference paper"
}