Title :
Neural network models for preprocessing and discriminating utterances of consonant-vowel units
Author :
Gangashetty, Suryakanth V. ; Khan, A. Nayeemulla ; Prasanna, S. R Mahadeva ; Yegnanarayana, B.
Author_Institution :
Dept. of Comput. Sci. & Eng., Indian Inst. of Technol., Madras, India
fDate :
6/24/1905 12:00:00 AM
Abstract :
We demonstrate the significance of nonlinear neural network models for compression of feature vectors and also develop classifiers for syllable-like units. We consider the standard 80 stop consonant-vowel units of most Indian languages. This set consists of dynamic sounds and hence requires large size feature vectors to represent the acoustic characteristics of these units. To develop classifiers with limited training data, it is necessary to compress the size of the feature vector. We show that nonlinear compression by autoassociative neural network model is useful, and is superior to the compression by linear principal component analysis
Keywords :
data compression; feedforward neural nets; multilayer perceptrons; pattern classification; speech recognition; Indian languages; acoustic characteristics; autoassociative neural network model; consonant-vowel units; dynamic sounds; feature vectors compression; neural network models; nonlinear compression; nonlinear neural network models; syllable-like units; utterances discrimination; utterances preprocessing; Computer science; Laboratories; Multi-layer neural network; Natural languages; Neural networks; Principal component analysis; Production; Speech; Training data; Vectors;
Conference_Titel :
Neural Networks, 2002. IJCNN '02. Proceedings of the 2002 International Joint Conference on
Conference_Location :
Honolulu, HI
Print_ISBN :
0-7803-7278-6
DOI :
10.1109/IJCNN.2002.1005542
