A neural network using acoustic sub-word units for continuous speech recognition

Author

Yu, Ha-Jin ; Oh, Yung-Hwan

Author_Institution

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

Volume

1

fYear

1996

fDate

3-6 Oct 1996

Firstpage

506

Abstract

A subword-based neural network model for continuous speech recognition is proposed. The system consists of three modules, and each module is composed of simple neural networks. The speech input is segmented into non-uniform units by the network in the first module. Non-uniform unit can model phoneme variations which spread for several phonemes and between words. The second module recognizes segmented units. The unit has stationary and transition parts, and the network is divided according to the two parts. The last module spots words by modeling temporal representation. The results of speaker independent word spotting of 520 words are described

Keywords

feedforward neural nets; learning (artificial intelligence); speech recognition; acoustic sub-word units; continuous speech recognition; phoneme variations; segmented units; speaker independent word spotting; subword-based neural network model; temporal representation; Acoustic signal detection; Computer science; Frequency; Hidden Markov models; Humans; Neural network hardware; Neural networks; Parallel processing; Speech recognition; Speech synthesis;

fLanguage

English

Publisher

ieee

Conference_Titel

Spoken Language, 1996. ICSLP 96. Proceedings., Fourth International Conference on

Conference_Location

Philadelphia, PA

Print_ISBN

0-7803-3555-4

Type

conf

DOI

10.1109/ICSLP.1996.607165

Filename

607165