Affect recognition from speech

Li Zhang; Virginia Francisco

doi:10.3233/978-1-60750-028-5-683

Affect recognition from speech

Li Zhang, Virginia Francisco

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

31 Downloads (Pure)

Abstract

We aim to provide an automatic anti-bullying component in online text and speech based interaction for young people age 18 - 25. Affect expression in speech generally differs from culture to culture, from female to male. In this study, we focus on affect sensing from speech for different gender user groups for young people. So far our work mainly concentrates on the sensing of five basic emotions (including 'happiness', 'sadness', 'fear', 'surprise', and 'anger') and 'neutral' from speech. Detailed acoustic features have been extracted after analysis of speech data from one chosen male and female speaker. Our affect sensing component has been implemented under the theory of naïve Bayes classifier. We have also evaluated it using new test data. Our work contributes to the conference themes on intelligent technologies - machine learning and affective speech processing.

Original language	English
Title of host publication	Artificial intelligence in education - Building learning systems that care: From knowledge representation to affective modelling
Publisher	IOS Press
Pages	683 - 685
ISBN (Print)	9781607500285
DOIs	https://doi.org/10.3233/978-1-60750-028-5-683
Publication status	Published - 2009
Externally published	Yes

Bibliographical note

Subject to restrictions, author can archive publisher's version/PDF.

Access to Document

10.3233/978-1-60750-028-5-683

Full textLicence: CC BY-NC-ND

Cite this

@inproceedings{fab6505519c2469d9e70d990c7d34624,

title = "Affect recognition from speech",

abstract = "We aim to provide an automatic anti-bullying component in online text and speech based interaction for young people age 18 - 25. Affect expression in speech generally differs from culture to culture, from female to male. In this study, we focus on affect sensing from speech for different gender user groups for young people. So far our work mainly concentrates on the sensing of five basic emotions (including 'happiness', 'sadness', 'fear', 'surprise', and 'anger') and 'neutral' from speech. Detailed acoustic features have been extracted after analysis of speech data from one chosen male and female speaker. Our affect sensing component has been implemented under the theory of na{\"i}ve Bayes classifier. We have also evaluated it using new test data. Our work contributes to the conference themes on intelligent technologies - machine learning and affective speech processing.",

author = "Li Zhang and Virginia Francisco",

note = "Subject to restrictions, author can archive publisher's version/PDF. ",

year = "2009",

doi = "10.3233/978-1-60750-028-5-683",

language = "English",

isbn = "9781607500285",

pages = "683 -- 685",

booktitle = "Artificial intelligence in education - Building learning systems that care: From knowledge representation to affective modelling",

publisher = "IOS Press",

address = "Netherlands",

}

TY - GEN

T1 - Affect recognition from speech

AU - Zhang, Li

AU - Francisco, Virginia

N1 - Subject to restrictions, author can archive publisher's version/PDF.

PY - 2009

Y1 - 2009

N2 - We aim to provide an automatic anti-bullying component in online text and speech based interaction for young people age 18 - 25. Affect expression in speech generally differs from culture to culture, from female to male. In this study, we focus on affect sensing from speech for different gender user groups for young people. So far our work mainly concentrates on the sensing of five basic emotions (including 'happiness', 'sadness', 'fear', 'surprise', and 'anger') and 'neutral' from speech. Detailed acoustic features have been extracted after analysis of speech data from one chosen male and female speaker. Our affect sensing component has been implemented under the theory of naïve Bayes classifier. We have also evaluated it using new test data. Our work contributes to the conference themes on intelligent technologies - machine learning and affective speech processing.

AB - We aim to provide an automatic anti-bullying component in online text and speech based interaction for young people age 18 - 25. Affect expression in speech generally differs from culture to culture, from female to male. In this study, we focus on affect sensing from speech for different gender user groups for young people. So far our work mainly concentrates on the sensing of five basic emotions (including 'happiness', 'sadness', 'fear', 'surprise', and 'anger') and 'neutral' from speech. Detailed acoustic features have been extracted after analysis of speech data from one chosen male and female speaker. Our affect sensing component has been implemented under the theory of naïve Bayes classifier. We have also evaluated it using new test data. Our work contributes to the conference themes on intelligent technologies - machine learning and affective speech processing.

U2 - 10.3233/978-1-60750-028-5-683

DO - 10.3233/978-1-60750-028-5-683

M3 - Conference contribution

SN - 9781607500285

SP - 683

EP - 685

BT - Artificial intelligence in education - Building learning systems that care: From knowledge representation to affective modelling

PB - IOS Press

ER -

Affect recognition from speech

Abstract

Bibliographical note

Access to Document

Fingerprint

Cite this