The recognition of isolated words on a speaker dependent system

Author

Wong, Choi Wan ; Mosttafavi, M.T.

Author_Institution

Dept. of Comput. Sci., North Carolina Univ., Charlotte, NC, USA

fYear

1989

fDate

9-12 Apr 1989

Firstpage

755

Abstract

The speaker-dependent recognition of isolated words is done on a MC68000 microprocessor system. Templates are made after performing data selections and compressions on vocabularies. The input pattern obtained during run time is classified according to the zero crossing rate and the number of matching elements. It is concluded that the speech recognition system faces two major problems in comparison and averaging of templates. Speed becomes more and more critical when the size of vocabularies increases. Distortion may result during template averaging. These problems may be tackled by combining zero crossing rates and template matching methods into a two-phased recognition strategy. Template matching is performed on candidates with close zero crossing values. This may speed up the process as well as ensure the accuracy of the system

Keywords

speech recognition; MC68000 microprocessor system; data compression; data selections; distortion; isolated words recognition; matching elements; speaker dependent recognition system; speech recognition system; template averaging; template matching methods; vocabularies; zero crossing rate; Band pass filters; Bandwidth; Delay effects; Filtering theory; Frequency; Microprocessors; Prototypes; Speech analysis; Speech recognition; Vocabulary;

fLanguage

English

Publisher

ieee

Conference_Titel

Southeastcon '89. Proceedings. Energy and Information Technologies in the Southeast., IEEE

Conference_Location

Columbia, SC

Type

conf

DOI

10.1109/SECON.1989.132501

Filename

132501