Deep Learning for the Detection of Emotion in Human Speech: The Impact of Audio Sample Duration and English versus Italian Languages

Alexander Wurst; Michael Hopwood; Sifan Wu; Fei Li; Yu Dong Yao

doi:10.1109/WOCC58016.2023.10139686

Deep Learning for the Detection of Emotion in Human Speech: The Impact of Audio Sample Duration and English versus Italian Languages

Alexander Wurst
, Michael Hopwood
, Sifan Wu
, Fei Li
, Yu Dong Yao

Department of Electrical and Computer Engineering

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

5 Scopus citations

Abstract

Identification of emotion types is important in the diagnosis and treatment of certain mental illnesses. This study uses audio data and deep learning methods such as convolutional neural networks (CNN) and long short-term memory (LSTM) to classify the emotion of human speech. We use the IEMOCAP and DEMoS datasets, consisting of English and Italian audio speech data in our experiments to classify speech into one of up to four emotions: angry, happy, neutral, and sad. The classification performance results demonstrate the effectiveness of the deep learning methods and our experiments yield between 62 and 92 percent classification accuracies. We specifically investigate the impact of the audio sample duration on the classification accuracy. In addition, we examine and compare the classification accuracy for English versus Italian languages.

Original language	English
Title of host publication	32nd Wireless and Optical Communications Conference, WOCC 2023
ISBN (Electronic)	9798350337150
DOIs	https://doi.org/10.1109/WOCC58016.2023.10139686
State	Published - 2023
Event	32nd Wireless and Optical Communications Conference, WOCC 2023 - Newark, United States Duration: 5 May 2023 → 6 May 2023

Publication series

Name	32nd Wireless and Optical Communications Conference, WOCC 2023

Conference

Conference	32nd Wireless and Optical Communications Conference, WOCC 2023
Country/Territory	United States
City	Newark
Period	5/05/23 → 6/05/23

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

convolutional neural network (CNN)
deep learning
emotion recognition
long short-term memory (LSTM)
spectrogram

Access to Document

10.1109/WOCC58016.2023.10139686

Cite this

Wurst, A., Hopwood, M., Wu, S., Li, F., & Yao, Y. D. (2023). Deep Learning for the Detection of Emotion in Human Speech: The Impact of Audio Sample Duration and English versus Italian Languages. In 32nd Wireless and Optical Communications Conference, WOCC 2023 (32nd Wireless and Optical Communications Conference, WOCC 2023). https://doi.org/10.1109/WOCC58016.2023.10139686

@inproceedings{0cdfd8c935bc4bb0ada5e70339b32a33,

title = "Deep Learning for the Detection of Emotion in Human Speech: The Impact of Audio Sample Duration and English versus Italian Languages",

abstract = "Identification of emotion types is important in the diagnosis and treatment of certain mental illnesses. This study uses audio data and deep learning methods such as convolutional neural networks (CNN) and long short-term memory (LSTM) to classify the emotion of human speech. We use the IEMOCAP and DEMoS datasets, consisting of English and Italian audio speech data in our experiments to classify speech into one of up to four emotions: angry, happy, neutral, and sad. The classification performance results demonstrate the effectiveness of the deep learning methods and our experiments yield between 62 and 92 percent classification accuracies. We specifically investigate the impact of the audio sample duration on the classification accuracy. In addition, we examine and compare the classification accuracy for English versus Italian languages.",

keywords = "convolutional neural network (CNN), deep learning, emotion recognition, long short-term memory (LSTM), spectrogram",

author = "Alexander Wurst and Michael Hopwood and Sifan Wu and Fei Li and Yao, \{Yu Dong\}",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 32nd Wireless and Optical Communications Conference, WOCC 2023 ; Conference date: 05-05-2023 Through 06-05-2023",

year = "2023",

doi = "10.1109/WOCC58016.2023.10139686",

language = "English",

series = "32nd Wireless and Optical Communications Conference, WOCC 2023",

booktitle = "32nd Wireless and Optical Communications Conference, WOCC 2023",

}

Wurst, A, Hopwood, M, Wu, S, Li, F & Yao, YD 2023, Deep Learning for the Detection of Emotion in Human Speech: The Impact of Audio Sample Duration and English versus Italian Languages. in 32nd Wireless and Optical Communications Conference, WOCC 2023. 32nd Wireless and Optical Communications Conference, WOCC 2023, 32nd Wireless and Optical Communications Conference, WOCC 2023, Newark, United States, 5/05/23. https://doi.org/10.1109/WOCC58016.2023.10139686

Deep Learning for the Detection of Emotion in Human Speech: The Impact of Audio Sample Duration and English versus Italian Languages. / Wurst, Alexander; Hopwood, Michael; Wu, Sifan et al.
32nd Wireless and Optical Communications Conference, WOCC 2023. 2023. (32nd Wireless and Optical Communications Conference, WOCC 2023).

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

TY - GEN

T1 - Deep Learning for the Detection of Emotion in Human Speech

T2 - 32nd Wireless and Optical Communications Conference, WOCC 2023

AU - Wurst, Alexander

AU - Hopwood, Michael

AU - Wu, Sifan

AU - Li, Fei

AU - Yao, Yu Dong

PY - 2023

Y1 - 2023

N2 - Identification of emotion types is important in the diagnosis and treatment of certain mental illnesses. This study uses audio data and deep learning methods such as convolutional neural networks (CNN) and long short-term memory (LSTM) to classify the emotion of human speech. We use the IEMOCAP and DEMoS datasets, consisting of English and Italian audio speech data in our experiments to classify speech into one of up to four emotions: angry, happy, neutral, and sad. The classification performance results demonstrate the effectiveness of the deep learning methods and our experiments yield between 62 and 92 percent classification accuracies. We specifically investigate the impact of the audio sample duration on the classification accuracy. In addition, we examine and compare the classification accuracy for English versus Italian languages.

AB - Identification of emotion types is important in the diagnosis and treatment of certain mental illnesses. This study uses audio data and deep learning methods such as convolutional neural networks (CNN) and long short-term memory (LSTM) to classify the emotion of human speech. We use the IEMOCAP and DEMoS datasets, consisting of English and Italian audio speech data in our experiments to classify speech into one of up to four emotions: angry, happy, neutral, and sad. The classification performance results demonstrate the effectiveness of the deep learning methods and our experiments yield between 62 and 92 percent classification accuracies. We specifically investigate the impact of the audio sample duration on the classification accuracy. In addition, we examine and compare the classification accuracy for English versus Italian languages.

KW - convolutional neural network (CNN)

KW - deep learning

KW - emotion recognition

KW - long short-term memory (LSTM)

KW - spectrogram

UR - https://www.scopus.com/pages/publications/85162716669

UR - https://www.scopus.com/pages/publications/85162716669#tab=citedBy

U2 - 10.1109/WOCC58016.2023.10139686

DO - 10.1109/WOCC58016.2023.10139686

M3 - Conference contribution

AN - SCOPUS:85162716669

T3 - 32nd Wireless and Optical Communications Conference, WOCC 2023

BT - 32nd Wireless and Optical Communications Conference, WOCC 2023

Y2 - 5 May 2023 through 6 May 2023

ER -

Deep Learning for the Detection of Emotion in Human Speech: The Impact of Audio Sample Duration and English versus Italian Languages

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this