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

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F18%3APU130758" target="_blank" >RIV/00216305:26230/18:PU130758 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.isca-speech.org/archive/Odyssey_2018/pdfs/42.pdf" target="_blank" >https://www.isca-speech.org/archive/Odyssey_2018/pdfs/42.pdf</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.21437/Odyssey.2018-6" target="_blank" >10.21437/Odyssey.2018-6</a>

Jazyk výsledku

angličtina

Název v původním jazyce

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

Popis výsledku v původním jazyce

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.

Název v anglickém jazyce

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

Popis výsledku anglicky

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.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

Výsledek vznikl pri realizaci vícero projektů. Více informací v záložce Projekty.

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2018

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of Odyssey 2018 The Speaker and Language Recognition Workshop

ISBN

—

ISSN

2312-2846

e-ISSN

—

Počet stran výsledku

8

Strana od-do

39-46

Název nakladatele

International Speech Communication Association

Místo vydání

Les Sables d'Olonne

Místo konání akce

Les Sables d'Olonne, France

Datum konání akce

26. 6. 2018

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—