HMM-based acoustic event detection with adaboost feature selection

Xi Zhou, Xiaodan Zhuang, Ming Liu, Hao Tang, Mark Allan Hasegawa-Johnson, Thomas S Huang

Research output: Contribution to journalConference article

Abstract

Because of the spectral difference between speech and acous- tic events, we propose using Kullback-Leibler distance to quantify the discriminant capability of all speech feature components in acoustic event detection. Based on these distances, we use AdaBoost to select a discriminant feature set and demonstrate that this feature set outperforms classical speech feature set such as MFCC in one-pass HMM-based acoustic event detection. We implement an HMM-based acoustic events detection system with lattice rescoring using a feature set selected by the above AdaBoost based approach.

Original languageEnglish (US)
Pages (from-to)345-353
Number of pages9
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4625 LNCS
DOIs
StatePublished - Jul 28 2008
Event2nd Annual Classifcation of Events Activities and Relationships, CLEAR 2007 and Rich Transcription, RT 2007 - Baltimore, MD, United States
Duration: May 8 2007May 11 2007

Fingerprint

Adaptive boosting
Event Detection
AdaBoost
Feature Selection
Feature extraction
Acoustics
Discriminant
Kullback-Leibler Distance
Quantify
Demonstrate
Speech

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

@article{6fe31143e19c4671bc88093ca503f166,
title = "HMM-based acoustic event detection with adaboost feature selection",
abstract = "Because of the spectral difference between speech and acous- tic events, we propose using Kullback-Leibler distance to quantify the discriminant capability of all speech feature components in acoustic event detection. Based on these distances, we use AdaBoost to select a discriminant feature set and demonstrate that this feature set outperforms classical speech feature set such as MFCC in one-pass HMM-based acoustic event detection. We implement an HMM-based acoustic events detection system with lattice rescoring using a feature set selected by the above AdaBoost based approach.",
author = "Xi Zhou and Xiaodan Zhuang and Ming Liu and Hao Tang and Hasegawa-Johnson, {Mark Allan} and Huang, {Thomas S}",
year = "2008",
month = "7",
day = "28",
doi = "10.1007/978-3-540-68585-2_33",
language = "English (US)",
volume = "4625 LNCS",
pages = "345--353",
journal = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
issn = "0302-9743",
publisher = "Springer Verlag",

}

TY - JOUR

T1 - HMM-based acoustic event detection with adaboost feature selection

AU - Zhou, Xi

AU - Zhuang, Xiaodan

AU - Liu, Ming

AU - Tang, Hao

AU - Hasegawa-Johnson, Mark Allan

AU - Huang, Thomas S

PY - 2008/7/28

Y1 - 2008/7/28

N2 - Because of the spectral difference between speech and acous- tic events, we propose using Kullback-Leibler distance to quantify the discriminant capability of all speech feature components in acoustic event detection. Based on these distances, we use AdaBoost to select a discriminant feature set and demonstrate that this feature set outperforms classical speech feature set such as MFCC in one-pass HMM-based acoustic event detection. We implement an HMM-based acoustic events detection system with lattice rescoring using a feature set selected by the above AdaBoost based approach.

AB - Because of the spectral difference between speech and acous- tic events, we propose using Kullback-Leibler distance to quantify the discriminant capability of all speech feature components in acoustic event detection. Based on these distances, we use AdaBoost to select a discriminant feature set and demonstrate that this feature set outperforms classical speech feature set such as MFCC in one-pass HMM-based acoustic event detection. We implement an HMM-based acoustic events detection system with lattice rescoring using a feature set selected by the above AdaBoost based approach.

UR - http://www.scopus.com/inward/record.url?scp=47749116035&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=47749116035&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-68585-2_33

DO - 10.1007/978-3-540-68585-2_33

M3 - Conference article

VL - 4625 LNCS

SP - 345

EP - 353

JO - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

JF - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SN - 0302-9743

ER -