Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information

Bagus Tris Atmaja; Masato Akagi

doi:10.1109/O-COCOSDA50338.2020.9295032

Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information

Bagus Tris Atmaja, Masato Akagi

Laboratory of Vibration and Acoustic

Japan Advanced Institute of Science and Technology

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

6 Citations (Scopus)

Abstract

In dimensional emotion recognition, a model called valence, arousal, and dominance is widely used. The current research in dimensional speech emotion recognition has shown a problem that the performance of valence prediction is lower than arousal and dominance. This paper presents an approach to tackle this problem: Improving the low score of valence prediction by utilizing linguistic information. Our approach fuses acoustic features with linguistic features, which is a conversion from words to vectors. The results doubled the performance of valence prediction on both single-task learning single-output (predicting valence only) and multitask learning multi-output (predicting valence, arousal, and dominance). Using a proper combination of acoustic and linguistic features not only improved valence prediction, but also improved arousal and dominance predictions in multitask learning.

Original language	English
Title of host publication	Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	166-171
Number of pages	6
ISBN (Electronic)	9781728198965
DOIs	https://doi.org/10.1109/O-COCOSDA50338.2020.9295032
Publication status	Published - 5 Nov 2020
Event	23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020 - Virtual, Yangon, Myanmar Duration: 5 Nov 2020 → 7 Nov 2020

Publication series

Name	Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020

Conference

Conference	23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020
Country/Territory	Myanmar
City	Virtual, Yangon
Period	5/11/20 → 7/11/20

Keywords

affective computing
dimensional emotion
linguistic feature
speech emotion recognition
valence prediction

Access to Document

10.1109/O-COCOSDA50338.2020.9295032

Cite this

Atmaja, B. T., & Akagi, M. (2020). Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information. In Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020 (pp. 166-171). Article 9295032 (Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/O-COCOSDA50338.2020.9295032

Atmaja, Bagus Tris ; Akagi, Masato. / Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information. Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020. Institute of Electrical and Electronics Engineers Inc., 2020. pp. 166-171 (Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020).

@inproceedings{aa73b535a9e1416d93df2edf2bca003c,

title = "Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information",

abstract = "In dimensional emotion recognition, a model called valence, arousal, and dominance is widely used. The current research in dimensional speech emotion recognition has shown a problem that the performance of valence prediction is lower than arousal and dominance. This paper presents an approach to tackle this problem: Improving the low score of valence prediction by utilizing linguistic information. Our approach fuses acoustic features with linguistic features, which is a conversion from words to vectors. The results doubled the performance of valence prediction on both single-task learning single-output (predicting valence only) and multitask learning multi-output (predicting valence, arousal, and dominance). Using a proper combination of acoustic and linguistic features not only improved valence prediction, but also improved arousal and dominance predictions in multitask learning.",

keywords = "affective computing, dimensional emotion, linguistic feature, speech emotion recognition, valence prediction",

author = "Atmaja, {Bagus Tris} and Masato Akagi",

note = "Publisher Copyright: {\textcopyright} 2020 IEEE.; 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020 ; Conference date: 05-11-2020 Through 07-11-2020",

year = "2020",

month = nov,

day = "5",

doi = "10.1109/O-COCOSDA50338.2020.9295032",

language = "English",

series = "Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "166--171",

booktitle = "Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020",

address = "United States",

}

Atmaja, BT & Akagi, M 2020, Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information. in Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020., 9295032, Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020, Institute of Electrical and Electronics Engineers Inc., pp. 166-171, 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020, Virtual, Yangon, Myanmar, 5/11/20. https://doi.org/10.1109/O-COCOSDA50338.2020.9295032

Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information. / Atmaja, Bagus Tris; Akagi, Masato.
Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020. Institute of Electrical and Electronics Engineers Inc., 2020. p. 166-171 9295032 (Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020).

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

TY - GEN

T1 - Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information

AU - Atmaja, Bagus Tris

AU - Akagi, Masato

PY - 2020/11/5

Y1 - 2020/11/5

N2 - In dimensional emotion recognition, a model called valence, arousal, and dominance is widely used. The current research in dimensional speech emotion recognition has shown a problem that the performance of valence prediction is lower than arousal and dominance. This paper presents an approach to tackle this problem: Improving the low score of valence prediction by utilizing linguistic information. Our approach fuses acoustic features with linguistic features, which is a conversion from words to vectors. The results doubled the performance of valence prediction on both single-task learning single-output (predicting valence only) and multitask learning multi-output (predicting valence, arousal, and dominance). Using a proper combination of acoustic and linguistic features not only improved valence prediction, but also improved arousal and dominance predictions in multitask learning.

AB - In dimensional emotion recognition, a model called valence, arousal, and dominance is widely used. The current research in dimensional speech emotion recognition has shown a problem that the performance of valence prediction is lower than arousal and dominance. This paper presents an approach to tackle this problem: Improving the low score of valence prediction by utilizing linguistic information. Our approach fuses acoustic features with linguistic features, which is a conversion from words to vectors. The results doubled the performance of valence prediction on both single-task learning single-output (predicting valence only) and multitask learning multi-output (predicting valence, arousal, and dominance). Using a proper combination of acoustic and linguistic features not only improved valence prediction, but also improved arousal and dominance predictions in multitask learning.

KW - affective computing

KW - dimensional emotion

KW - linguistic feature

KW - speech emotion recognition

KW - valence prediction

UR - http://www.scopus.com/inward/record.url?scp=85099552107&partnerID=8YFLogxK

U2 - 10.1109/O-COCOSDA50338.2020.9295032

DO - 10.1109/O-COCOSDA50338.2020.9295032

M3 - Conference contribution

AN - SCOPUS:85099552107

T3 - Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020

SP - 166

EP - 171

BT - Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020

Y2 - 5 November 2020 through 7 November 2020

ER -

Atmaja BT, Akagi M. Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information. In Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020. Institute of Electrical and Electronics Engineers Inc. 2020. p. 166-171. 9295032. (Proceedings of 2020 23rd Conference of the Oriental COCOSDA International Committee for the Co-Ordination and Standardisation of Speech Databases and Assessment Techniques, O-COCOSDA 2020). doi: 10.1109/O-COCOSDA50338.2020.9295032

Improving Valence Prediction in Dimensional Speech Emotion Recognition Using Linguistic Information

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this