Publication detail

An Automatic Emotion Recognizer using MFCCs and Hidden Markov Models

VYAS, G. DUTTA, M. ŘÍHA, K. PŘINOSIL, J.

Original Title

An Automatic Emotion Recognizer using MFCCs and Hidden Markov Models

English Title

An Automatic Emotion Recognizer using MFCCs and Hidden Markov Models

Type

conference paper

Language

en

Original Abstract

In this paper, the proficiency of continuous Hidden Markov Models to recognize emotions from speech signals has been investigated. Unlike the existing work which considers prosodic features for automatic emotion recognition, this work proposes the effectiveness of the phonetic features of speech particularly, Mel-Frequency Cepstral Coefficients which improves the accuracy with reduced feature set. The continuous speech emotional utterances used in this work have been taken from the SAVEE emotional corpus. The Hidden Markov Model Toolkit (HTK) version 3.4.1 was utilized for extraction of the acoustic features as well as generation of the models. Optimizing the acoustic and pre-processing parameters along with the number of states and transition probabilities of the Markov Models, the trials give us an average accuracy of 78% and highest accuracy of 91.25% for four emotions sadness, surprise, fear and disgust.

English abstract

In this paper, the proficiency of continuous Hidden Markov Models to recognize emotions from speech signals has been investigated. Unlike the existing work which considers prosodic features for automatic emotion recognition, this work proposes the effectiveness of the phonetic features of speech particularly, Mel-Frequency Cepstral Coefficients which improves the accuracy with reduced feature set. The continuous speech emotional utterances used in this work have been taken from the SAVEE emotional corpus. The Hidden Markov Model Toolkit (HTK) version 3.4.1 was utilized for extraction of the acoustic features as well as generation of the models. Optimizing the acoustic and pre-processing parameters along with the number of states and transition probabilities of the Markov Models, the trials give us an average accuracy of 78% and highest accuracy of 91.25% for four emotions sadness, surprise, fear and disgust.

Keywords

Emotion; recognition; HTK toolkit; Mel frequency cepstral coefficients

RIV year

2015

Released

08.10.2015

Location

Brno, Czech Republic

ISBN

978-1-4673-9282-2

Book

2015 7th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT)

Pages from

320

Pages to

324

Pages count

5

BibTex


@inproceedings{BUT117824,
  author="Garima {Vyas} and Malay Kishore {Dutta} and Kamil {Říha} and Jiří {Přinosil}",
  title="An Automatic Emotion Recognizer using MFCCs and Hidden Markov Models",
  annote="In this paper, the proficiency of continuous Hidden Markov Models to recognize emotions from speech signals has been investigated. Unlike the existing work which considers prosodic features for automatic emotion recognition, this work proposes the effectiveness of the phonetic features of speech particularly, Mel-Frequency Cepstral Coefficients which improves the accuracy with reduced feature set. The continuous speech emotional utterances used in this work have been taken from the SAVEE emotional corpus. The Hidden Markov Model Toolkit (HTK) version 3.4.1 was utilized for extraction of the acoustic features as well as generation of the models. Optimizing the acoustic and pre-processing parameters along with the number of states and transition probabilities of the Markov Models, the trials give us an average accuracy of 78% and highest accuracy of 91.25% for four emotions sadness, surprise, fear and disgust.",
  booktitle="2015 7th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT)",
  chapter="117824",
  howpublished="electronic, physical medium",
  year="2015",
  month="october",
  pages="320--324",
  type="conference paper"
}