Faculty of Information Technology, BUT

Publication Details

Multilingually Trained Bottleneck Features in Spoken Language Recognition

FÉR Radek, MATĚJKA Pavel, GRÉZL František, PLCHOT Oldřich, VESELÝ Karel and ČERNOCKÝ Jan. Multilingually Trained Bottleneck Features in Spoken Language Recognition. Computer Speech and Language, vol. 2017, no. 46, pp. 252-267. ISSN 0885-2308. Available from: http://www.sciencedirect.com/science/article/pii/S0885230816302947
Czech title
Vícejazyčně trénované parametry založené na úzkém hrdle neuronových sítí pro rozpoznávání mluveného jazyka
Type
journal article
Language
english
Authors
URL
Keywords
Multilingual training, Bottleneck features, Spoken language recognition
Abstract
Multilingual training of neural networks has proven to be simple yet effective way to deal with multilingual training corpora. It allows to use several resources to jointly train a language independent representation of features, which can be encoded into low-dimensional feature set by embedding narrow bottleneck layer to the network. In this paper, we analyze such features on the task of spoken language recognition (SLR), focusing on practical aspects of training bottleneck networks and analyzing their integration in SLR. By comparing properties of mono and multilingual features we show the suitability of multilingual training for SLR. The state-of-the-art performance of these features is demonstrated on the NIST LRE09 database.
Annotation
Multilingual training of neural networks has proven to be simple yet effective way to deal with multilingual training corpora. It allows to use several resources to jointly train a language independent representation of features, which can be encoded into low-dimensional feature set by embedding narrow bottleneck layer to the network. In this paper, we analyze such features on the task of spoken language recognition (SLR), focusing on practical aspects of training bottleneck networks and analyzing their integration in SLR. By comparing properties of mono and multilingual features we show the suitability of multilingual training for SLR. The state-of-the-art performance of these features is demonstrated on the NIST LRE09 database.
Published
2017
Pages
252-267
Journal
Computer Speech and Language, vol. 2017, no. 46, ISSN 0885-2308
Publisher
Elsevier Science
DOI
BibTeX
@ARTICLE{FITPUB11518,
   author = "Radek F\'{e}r and Pavel Mat\v{e}jka and Franti\v{s}ek Gr\'{e}zl and Old\v{r}ich Plchot and Karel Vesel\'{y} and Jan \v{C}ernock\'{y}",
   title = "Multilingually Trained Bottleneck Features in Spoken Language Recognition",
   pages = "252--267",
   journal = "Computer Speech and Language",
   volume = 2017,
   number = 46,
   year = 2017,
   ISSN = "0885-2308",
   doi = "10.1016/j.csl.2017.06.008",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11518"
}
Back to top