Faculty of Information Technology, BUT

Publication Details

Sequence Summarizing Neural Networks for Spoken Language Recognition

PEŠÁN Jan, BURGET Lukáš and ČERNOCKÝ Jan. Sequence Summarizing Neural Networks for Spoken Language Recognition. In: Proceedings of Interspeech 2016. San Francisco: International Speech Communication Association, 2016, pp. 3285-3289. ISBN 978-1-5108-3313-5. Available from: https://www.researchgate.net/publication/307889421_Sequence_Summarizing_Neural_Networks_for_Spoken_Language_Recognition
Czech title
Sekvenční sumarizační neuronové sítě pro rozpoznávání mluveného jazyka
Type
conference paper
Language
english
Authors
URL
Keywords
Sequence Summarizing Neural Network, DNN, i-vectors
Abstract
This paper explores the use of Sequence Summarizing Neural Networks (SSNNs) as a variant of deep neural networks (DNNs) for classifying sequences. In this work, it is applied to the task of spoken language recognition. Unlike other classification tasks in speech processing where the DNN needs to produce a per-frame output, language is considered constant during an utterance. We introduce a summarization component into the DNN structure producing one set of language posteriors per utterance. The training of the DNN is performed by an appropriately modified gradient-descent algorithm. In our initial experiments, the SSNN results are compared to a single state-of-the-art i-vector based baseline system with a similar complexity (i.e. no system fusion, etc.). For some conditions, SSNNs is able to provide performance comparable to the baseline system. Relative improvement up to 30% is obtained with the score level fusion of the baseline and the SSNN systems.
Annotation
This paper explores the use of Sequence Summarizing Neural Networks (SSNNs) as a variant of deep neural networks (DNNs) for classifying sequences. In this work, it is applied to the task of spoken language recognition. Unlike other classification tasks in speech processing where the DNN needs to produce a per-frame output, language is considered constant during an utterance. We introduce a summarization component into the DNN structure producing one set of language posteriors per utterance. The training of the DNN is performed by an appropriately modified gradient-descent algorithm. In our initial experiments, the SSNN results are compared to a single state-of-the-art i-vector based baseline system with a similar complexity (i.e. no system fusion, etc.). For some conditions, SSNNs is able to provide performance comparable to the baseline system. Relative improvement up to 30% is obtained with the score level fusion of the baseline and the SSNN systems.
Published
2016
Pages
3285-3289
Proceedings
Proceedings of Interspeech 2016
Conference
Interspeech 2016, San Francisco, US
ISBN
978-1-5108-3313-5
Publisher
International Speech Communication Association
Place
San Francisco, US
DOI
BibTeX
@INPROCEEDINGS{FITPUB11273,
   author = "Jan Pe\v{s}\'{a}n and Luk\'{a}\v{s} Burget and Jan \v{C}ernock\'{y}",
   title = "Sequence Summarizing Neural Networks for Spoken Language Recognition",
   pages = "3285--3289",
   booktitle = "Proceedings of Interspeech 2016",
   year = 2016,
   location = "San Francisco, US",
   publisher = "International Speech Communication Association",
   ISBN = "978-1-5108-3313-5",
   doi = "10.21437/Interspeech.2016-764",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11273"
}
Files
Back to top