Fast decoder design of connected word speech recognition for automobile navigation system

Author

Jeong, Seokyeong ; Min, Kyoung Won ; Ko, Hanseok

Author_Institution

Dept. of Visual Inf. Process., Korea Univ., Seoul, South Korea

fYear

2006

fDate

7-11 Jan. 2006

Firstpage

215

Lastpage

216

Abstract

This paper presents efficient decoder techniques for use in a large vocabulary connected word speech recognition system that is designed to operate in an automobile navigation platform. In particular, we introduce a two-pass search technique that can handle a large vocabulary list at a high speed. Essentially, both the 1st and 2nd passes share the identical acoustic model that makes it possible to significantly reduce the amount of computational load in the 2nd pass. Also, efficient implementation techniques for embedded platforms are presented for realizing a high-speed speech recognizer design suitable for automobile navigation system.

Keywords

automobiles; decoding; navigation; speech coding; speech recognition; vocabulary; acoustic model; automobile navigation system; fast decoder design; high-speed speech recognizer; two-pass search technique; vocabulary connection; word speech recognition; Automobiles; Decoding; EMP radiation effects; Electrons; Hidden Markov models; Ice; Navigation; Speech recognition; Tellurium; Tin; automobile navigation system; embedded system; fast decoder design; speech recognition; two-pass search;

fLanguage

English

Publisher

ieee

Conference_Titel

Consumer Electronics, 2006. ICCE '06. 2006 Digest of Technical Papers. International Conference on

Print_ISBN

0-7803-9459-3

Type

conf

DOI

10.1109/ICCE.2006.1598387

Filename

1598387