Study on Algorithms of Keyword Confusion Network Generation

DOI

10.2991/jcis.2008.43How to use a DOI?

Keywords

keyword spotting, confusion network, speech recognition

Abstract

Keyword spotting based on large vocabulary continuous speech recognition (LVCSR) is the main researching direction of keyword spotting field. Lattice as the middle result of LVCSR, is often used in this system. But because of its big size, the performance is not efficient as we expect to be. In this paper, lattice was optimized by confusion network (CN) to achieve higher recall rate and lower error rate. And the study on algorithms of keyword confusion network generation was expanded. In the system, the proposed algorithm increased the recall rate of confusion network to 87.11%, while it was 65.46% with N-best.

Copyright

© 2008, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Lei Zhang
AU  - Meimei Jia
AU  - Lili Guo
PY  - 2008/12
DA  - 2008/12
TI  - Study on Algorithms of Keyword Confusion Network Generation
BT  - Proceedings of the 11th Joint Conference on Information Sciences (JCIS 2008)
PB  - Atlantis Press
SP  - 250
EP  - 255
SN  - 1951-6851
UR  - https://doi.org/10.2991/jcis.2008.43
DO  - 10.2991/jcis.2008.43
ID  - Zhang2008/12
ER  -