Adaptive stream reliability modeling based on local dispersion measures for audio visual speech recognition

Lei Xie; Rong Chun Zhao; Zhi Qiang Liu

Adaptive stream reliability modeling based on local dispersion measures for audio visual speech recognition

Lei Xie, Rong Chun Zhao, Zhi Qiang Liu

School of Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

This paper proposes an adaptive stream reliability modeling technique for audio visual speech recognition (AVSR). As recognition conditions vary locally, we present two local measures - frame and window dispersions to depict the temporal discriminative powers and noise levels of both audio and visual streams. The dispersions are subsequently mapped to stream exponents according to the minimum classification error (MCE) criterion. Experiments on a connected-digits task show that our method consistently outperforms the popular Discriminative Training (DT) and Grid Search (GS) methods at various signal noise ratios (SNRs), improving for example word accuracy rate (WAR) from 94.7% to 96.4% at 28dB SNR.

Original language	English
Title of host publication	2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005
Pages	4852-4857
Number of pages	6
State	Published - 2005
Event	International Conference on Machine Learning and Cybernetics, ICMLC 2005 - Guangzhou, China Duration: 18 Aug 2005 → 21 Aug 2005

Publication series

Name	2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005

Conference

Conference	International Conference on Machine Learning and Cybernetics, ICMLC 2005
Country/Territory	China
City	Guangzhou
Period	18/08/05 → 21/08/05

Keywords

Audio visual speech recognition
Dispersion
Lipreading
MCE-GPD
Stream exponents

Cite this

@inproceedings{e9b709b61adb48cbab1138fe2e40bc39,

title = "Adaptive stream reliability modeling based on local dispersion measures for audio visual speech recognition",

abstract = "This paper proposes an adaptive stream reliability modeling technique for audio visual speech recognition (AVSR). As recognition conditions vary locally, we present two local measures - frame and window dispersions to depict the temporal discriminative powers and noise levels of both audio and visual streams. The dispersions are subsequently mapped to stream exponents according to the minimum classification error (MCE) criterion. Experiments on a connected-digits task show that our method consistently outperforms the popular Discriminative Training (DT) and Grid Search (GS) methods at various signal noise ratios (SNRs), improving for example word accuracy rate (WAR) from 94.7% to 96.4% at 28dB SNR.",

keywords = "Audio visual speech recognition, Dispersion, Lipreading, MCE-GPD, Stream exponents",

author = "Lei Xie and Zhao, {Rong Chun} and Liu, {Zhi Qiang}",

year = "2005",

language = "英语",

isbn = "078039092X",

series = "2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005",

pages = "4852--4857",

booktitle = "2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005",

note = "International Conference on Machine Learning and Cybernetics, ICMLC 2005 ; Conference date: 18-08-2005 Through 21-08-2005",

}

Xie, L, Zhao, RC & Liu, ZQ 2005, Adaptive stream reliability modeling based on local dispersion measures for audio visual speech recognition. in 2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005. 2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005, pp. 4852-4857, International Conference on Machine Learning and Cybernetics, ICMLC 2005, Guangzhou, China, 18/08/05.

Adaptive stream reliability modeling based on local dispersion measures for audio visual speech recognition. / Xie, Lei; Zhao, Rong Chun; Liu, Zhi Qiang.
2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005. 2005. p. 4852-4857 (2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Adaptive stream reliability modeling based on local dispersion measures for audio visual speech recognition

AU - Xie, Lei

AU - Zhao, Rong Chun

AU - Liu, Zhi Qiang

PY - 2005

Y1 - 2005

N2 - This paper proposes an adaptive stream reliability modeling technique for audio visual speech recognition (AVSR). As recognition conditions vary locally, we present two local measures - frame and window dispersions to depict the temporal discriminative powers and noise levels of both audio and visual streams. The dispersions are subsequently mapped to stream exponents according to the minimum classification error (MCE) criterion. Experiments on a connected-digits task show that our method consistently outperforms the popular Discriminative Training (DT) and Grid Search (GS) methods at various signal noise ratios (SNRs), improving for example word accuracy rate (WAR) from 94.7% to 96.4% at 28dB SNR.

AB - This paper proposes an adaptive stream reliability modeling technique for audio visual speech recognition (AVSR). As recognition conditions vary locally, we present two local measures - frame and window dispersions to depict the temporal discriminative powers and noise levels of both audio and visual streams. The dispersions are subsequently mapped to stream exponents according to the minimum classification error (MCE) criterion. Experiments on a connected-digits task show that our method consistently outperforms the popular Discriminative Training (DT) and Grid Search (GS) methods at various signal noise ratios (SNRs), improving for example word accuracy rate (WAR) from 94.7% to 96.4% at 28dB SNR.

KW - Audio visual speech recognition

KW - Dispersion

KW - Lipreading

KW - MCE-GPD

KW - Stream exponents

UR - http://www.scopus.com/inward/record.url?scp=28444438538&partnerID=8YFLogxK

M3 - 会议稿件

AN - SCOPUS:28444438538

SN - 078039092X

SN - 9780780390928

T3 - 2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005

SP - 4852

EP - 4857

BT - 2005 International Conference on Machine Learning and Cybernetics, ICMLC 2005

T2 - International Conference on Machine Learning and Cybernetics, ICMLC 2005

Y2 - 18 August 2005 through 21 August 2005

ER -

Adaptive stream reliability modeling based on local dispersion measures for audio visual speech recognition

Abstract

Publication series

Conference

Keywords

Other files and links

Fingerprint

Cite this