IG-based feature extraction and compensation for emotion recognition from speech

Ze Jing Chuang, Chung-Hsien Wu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

This paper presents an approach to emotion recognition from speech signals. In this approach, the intonation groups (IGs) of the input speech signals are firstly extracted. The speech features in each selected intonation group are then extracted. With the assumption of linear mapping between feature spaces in different emotional states, a feature compensation approach is proposed to characterize the feature space with better discriminability among emotional states. The compensation vector with respect to each emotional state is estimated using the Minimum Classification Error (MCE) algorithm. The IG-based feature vectors compensated by the compensation vectors are used to train the Gaussian Mixture Models (GMMs) for each emotional state. The emotional state with the GMM having the maximal likelihood ratio is determined as the final output. The experimental result shows that IG-based feature extraction and compensation can obtain encouraging performance for emotion recognition.

Original languageEnglish
Title of host publicationAffective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings
Pages358-365
Number of pages8
Publication statusPublished - 2005 Dec 1
Event1st International Conference on ffective Computing and Intelligent Interaction, ACII 2005 - Beijing, China
Duration: 2005 Oct 222005 Oct 24

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3784 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other1st International Conference on ffective Computing and Intelligent Interaction, ACII 2005
CountryChina
CityBeijing
Period05-10-2205-10-24

Fingerprint

Emotion Recognition
Feature Extraction
Feature extraction
Speech Signal
Gaussian Mixture Model
Feature Space
Likelihood Ratio
Feature Vector
Emotion
Speech
Compensation and Redress
Output
Experimental Results

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Chuang, Z. J., & Wu, C-H. (2005). IG-based feature extraction and compensation for emotion recognition from speech. In Affective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings (pp. 358-365). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3784 LNCS).
Chuang, Ze Jing ; Wu, Chung-Hsien. / IG-based feature extraction and compensation for emotion recognition from speech. Affective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings. 2005. pp. 358-365 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{726e12be10cd41ff9d3a0c8fcad28a3d,
title = "IG-based feature extraction and compensation for emotion recognition from speech",
abstract = "This paper presents an approach to emotion recognition from speech signals. In this approach, the intonation groups (IGs) of the input speech signals are firstly extracted. The speech features in each selected intonation group are then extracted. With the assumption of linear mapping between feature spaces in different emotional states, a feature compensation approach is proposed to characterize the feature space with better discriminability among emotional states. The compensation vector with respect to each emotional state is estimated using the Minimum Classification Error (MCE) algorithm. The IG-based feature vectors compensated by the compensation vectors are used to train the Gaussian Mixture Models (GMMs) for each emotional state. The emotional state with the GMM having the maximal likelihood ratio is determined as the final output. The experimental result shows that IG-based feature extraction and compensation can obtain encouraging performance for emotion recognition.",
author = "Chuang, {Ze Jing} and Chung-Hsien Wu",
year = "2005",
month = "12",
day = "1",
language = "English",
isbn = "3540296212",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "358--365",
booktitle = "Affective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings",

}

Chuang, ZJ & Wu, C-H 2005, IG-based feature extraction and compensation for emotion recognition from speech. in Affective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3784 LNCS, pp. 358-365, 1st International Conference on ffective Computing and Intelligent Interaction, ACII 2005, Beijing, China, 05-10-22.

IG-based feature extraction and compensation for emotion recognition from speech. / Chuang, Ze Jing; Wu, Chung-Hsien.

Affective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings. 2005. p. 358-365 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3784 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - IG-based feature extraction and compensation for emotion recognition from speech

AU - Chuang, Ze Jing

AU - Wu, Chung-Hsien

PY - 2005/12/1

Y1 - 2005/12/1

N2 - This paper presents an approach to emotion recognition from speech signals. In this approach, the intonation groups (IGs) of the input speech signals are firstly extracted. The speech features in each selected intonation group are then extracted. With the assumption of linear mapping between feature spaces in different emotional states, a feature compensation approach is proposed to characterize the feature space with better discriminability among emotional states. The compensation vector with respect to each emotional state is estimated using the Minimum Classification Error (MCE) algorithm. The IG-based feature vectors compensated by the compensation vectors are used to train the Gaussian Mixture Models (GMMs) for each emotional state. The emotional state with the GMM having the maximal likelihood ratio is determined as the final output. The experimental result shows that IG-based feature extraction and compensation can obtain encouraging performance for emotion recognition.

AB - This paper presents an approach to emotion recognition from speech signals. In this approach, the intonation groups (IGs) of the input speech signals are firstly extracted. The speech features in each selected intonation group are then extracted. With the assumption of linear mapping between feature spaces in different emotional states, a feature compensation approach is proposed to characterize the feature space with better discriminability among emotional states. The compensation vector with respect to each emotional state is estimated using the Minimum Classification Error (MCE) algorithm. The IG-based feature vectors compensated by the compensation vectors are used to train the Gaussian Mixture Models (GMMs) for each emotional state. The emotional state with the GMM having the maximal likelihood ratio is determined as the final output. The experimental result shows that IG-based feature extraction and compensation can obtain encouraging performance for emotion recognition.

UR - http://www.scopus.com/inward/record.url?scp=33646809022&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33646809022&partnerID=8YFLogxK

M3 - Conference contribution

SN - 3540296212

SN - 9783540296218

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 358

EP - 365

BT - Affective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings

ER -

Chuang ZJ, Wu C-H. IG-based feature extraction and compensation for emotion recognition from speech. In Affective Computing and Intelligent Interaction - First International Conference, ACII 2005, Proceedings. 2005. p. 358-365. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).