A neural network for 500 vocabulary word spotting using acoustic sub-word units

Author

Yu, Ha-Jin ; Oh, Yung-Hwan

Author_Institution

Dept. of Comput. Sci., Korea Adv. Inst. of Sci. & Technol., Taejon, South Korea

Volume

4

fYear

1997

fDate

21-24 Apr 1997

Firstpage

3277

Abstract

A neural network model based on a non-uniform unit for speaker-independent continuous speech recognition is proposed. The functions of the neural network model include segmenting the input speech into sub-word units, classifying the units and detecting words, and each of them is implemented by a module. The recognition unit we propose can include an arbitrary number of phonemes in a unit, so that it can absorb co-articulation effects which spread for several phonemes. The unit classifier module separates the speech into stationary and transition parts and use different parameters for them. The word detector module can learn all the pronunciation variations in the training data. The system is evaluated on a subset of the TIMIT speech data

Keywords

Hebbian learning; acoustic signal detection; acoustic signal processing; modules; neural nets; pattern classification; speech processing; speech recognition; TIMIT speech data; acoustic subword units; coarticulation effects; input speech segmentation; neural network model; nonuniform unit; phonemes; pronunciation variations; recognition unit; speaker independent continuous speech recognition; supervised Hebbian learning; training data; unit classifier module; vocabulary; word detector module; word spotting; Computer science; Detectors; Hidden Markov models; Neural networks; Parallel processing; Robustness; Speech analysis; Speech recognition; Training data; Vocabulary;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1997. ICASSP-97., 1997 IEEE International Conference on

Conference_Location

Munich

ISSN

1520-6149

Print_ISBN

0-8186-7919-0

Type

conf

DOI

10.1109/ICASSP.1997.595493

Filename

595493