Authors: Jakovljević, Nikša
Mišković, Dragiša
Janev, Marko 
Sečujski, Milan
Delić, Vlado
Affiliations: Mathematical Institute of the Serbian Academy of Sciences and Arts 
Title: Comparison of linear discriminant analysis approaches in automatic speech recognition
Journal: Elektronika ir Elektrotechnika
Volume: 19
Issue: 7
First page: 76
Last page: 79
Issue Date: 27-Sep-2013
Rank: M22
ISSN: 1392-1215
DOI: 10.5755/j01.eee.19.7.5167
Speech recognition systems are commonly modelled by hidden Markov models with Gaussian mixture models as observation density functions. These models have a significant number of parameters, which usually leads to the problem of data sparsity, especially for under-resourced languages such as Serbian. One of the ways to overcome the problem of data sparsity is the reduction of the number of features. Linear discriminant analysis (LDA) and heteroscedastic LDA (HLDA) are two common ways to reduce the dimensionality in an automatic speech recognition task. The paper compares the properties of speech recognition systems for Serbian in which both techniques are applied with variable types of input features as well as the number of output features of (H)LDA. The best results are obtained in the case of HLDA with input vectors consisting of concatenations of feature vectors across 7 successive frames, where each feature vector contains 12 mel frequency cepstral coefficients (MFCCs) and normalized energy, and the number of output features is 32 or 35.
Keywords: Linear discriminant analysis | Speech recognition
Publisher: Kauno Technologijos Universitetas
Project: Development of Dialogue Systems for Serbian and Other South Slavic Languages 

Show full item record


checked on May 23, 2024

Page view(s)

checked on May 10, 2024

Google ScholarTM




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