DSpace
 

Researchspace >
General science, engineering & technology >
General science, engineering & technology >
General science, engineering & technology >

Please use this identifier to cite or link to this item: http://hdl.handle.net/10204/5565

Title: Continuous speech recognition with sparse coding
Authors: Smit, WJ
Barnard, E
Keywords: Sparse coding
Spike train
Speech recognition
Linear generative model
Issue Date: Apr-2009
Publisher: Elsevier
Citation: Smit, WJ and Barnard, E. 2009. Continuous speech recognition with sparse coding. Computer Speech and Language, vol. 23(2), pp 200-219
Abstract: Sparse coding is an efficient way of coding information. In a sparse code most of the code elements are zero; very few are active. Sparse codes are intended to correspond to the spike trains with which biological neurons communicate. In this article, we show how sparse codes can be used to do continuous speech recognition. We use the TIDIGITS dataset to illustrate the process. First a waveform is transformed into a spectrogram, and a sparse code for the spectrogram is found by means of a linear generative model. The spike train is classified by making use of a spike train model and dynamic programming. It is computationally expensive to find a sparse code. We use an iterative subset selection algorithm with quadratic programming for this process. This algorithm finds a sparse code in reasonable time if the input is limited to a fairly coarse spectral resolution. At this resolution, our system achieves a word error rate of 19%, whereas a system based on Hidden Markov Models achieves a word error rate of 15% at the same resolution.
Description: Copyright: Elsevier 2009. This is an ABSTRACT ONLY.
URI: http://www.sciencedirect.com/science/article/pii/S0885230808000375
http://hdl.handle.net/10204/5565
ISSN: 0885-2308
Appears in Collections:Human language technologies
General science, engineering & technology

Files in This Item:

File Description SizeFormat
Smit_2009_ABSTRACT ONLY.pdf79.12 kBAdobe PDFView/Open
View Statistics

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

 

Valid XHTML 1.0! DSpace Software Copyright © 2002-2010  Duraspace - Feedback