Conference paper

 
Veselý, K., Karafiát, M., Grézl, F.: Convolutive Bottleneck Network Features for LVCSR, In: Proceedings of ASRU 2011, Big Island, Hawaii, US, IEEESP, 2011, p. 42-47, ISBN 978-1-4673-0366-8
Publication language:english
Original title:Convolutive Bottleneck Network Features for LVCSR
Title (cs):Príznaky z konvolucní síte s úzkým hrdlem pro LVCSR
Pages:42-47
Proceedings:Proceedings of ASRU 2011
Conference:IEEE 2011 Workshop on Automatic Speech Recognition and Understanding
Place:Big Island, Hawaii, US
Year:2011
ISBN:978-1-4673-0366-8
Publisher:IEEE Signal Processing Society
URL:http://www.fit.vutbr.cz/research/groups/speech/publi/2011/vesely_asru2011_00042.pdf [PDF]
Keywords
Bottleneck features, Tandem LVCSR system, linear bottleneck, Convolutional Bottleneck Network
Annotation
Workshop Article about novel features for tandem LVCSR system, which are based on Convolutive Bottleneck Network. It extends the previous work on Universal Context network by using linear bottleneck and expansion to Convolutive Bottleneck Network,
so all the parameters are trained together.
Abstract
In this paper, we focus on improvements of the bottleneck ANN in a Tandem LVCSR system. First, the influence of training set size and the ANN size is evaluated. Second, a very positive effect of linear bottleneck is shown. Finally a Convolutive Bottleneck Network is proposed as extension of the current stateof- the-art Universal Context Network. The proposed training method leads to 5.5% relative reduction of WER, compared to the Universal Context ANN baseline. The relative improvement compared to the 5-layer single-bottleneck network is 17.7%. The dataset ctstrain07 composed of more than 2000 hours of English Conversational Telephone Speech was used for the experiments. The TNet toolkit with CUDA GPGPU implementation was used for fast training.
BibTeX:
@INPROCEEDINGS{
   author = {Karel Veselý and Martin Karafiát and Frantisek Grézl},
   title = {Convolutive Bottleneck Network Features for LVCSR},
   pages = {42--47},
   booktitle = {Proceedings of ASRU 2011},
   year = {2011},
   location = {Big Island, Hawaii, US},
   publisher = {IEEE Signal Processing Society},
   ISBN = {978-1-4673-0366-8},
   language = {english},
   url = {http://www.fit.vutbr.cz/research/view_pub.php?id=9763}
}