Speech emotion recognition using neural networks and support vector machines

Name: Nattapong Kurpukdee

LCSH: Kasetsart University -- Theses. M.Eng. (Information and Communication Technology for Embedded Systems) 2018

Classification :.LCCS: TK7882.S65

LCSH: Kasetsart University. -- Department of Electrical Engineering -- Theses

LCSH: Speech processing systems

LCSH: Neural networks (Computer science)

LCSH: Pattern recognition systems

LCSH: Machine learning

Abstract: Recently, in human-machine interaction (HMI), we are still far from being able to fully communicate with machines, partly because it is difficult for machines to interpret some paralinguistic information appeared in a spoken language like emotions. Speech emotion recognition (SER), which aims to classify speaker's emotional states through speech signals, is one of the essential tasks for making HMI more natural and realistic. Although SER has been widely studied and has attracted researchers' attention, the performance of SER systems developed so far remains relatively low, especially for spontaneous conversational speech. Consequently, improvement of SER performance is a crucial problem to be solved in HMI research area. Most researchers in SER have focused on feature extraction, feature selection and classification techniques using a frame-level. This study aims to classify speech signals into four emotional states (neutral, happiness, sadness, and anger). Exciting features have been derived recently by using a deep neural network (DNN). More recently, features given by recurrent neural network (RNN) and convolution neural network (CNN) have been proven to be efficient. Furthermore, both frame-level and utterance-level features have played their potential in speech emotion recognition. Therefore, this research takes all these advantages by using RNN, CNN, or their combination to extract frame-level phonemebased features to further construct utterance-level features. Afterward, Support Vector Machines (SVMs) and Linear discriminant analysis (LDA) are also utilized to classify emotional state.

Kasetsart University. Office of the University Library

Address: Bangkok

Email: tdckulib@ku.ac.th

Name: Poonlap Lamsrichan

Role: Thesis Advisor

Name: Chai Wutiwiwatchai

Role: Thesis CO-Advisor

Name: Kobayashi, Takao

Role: Thesis CO-Advisor

Created: 2018

Modified: 2568-07-05

Issued: 2025-07-05

วิทยานิพนธ์/Thesis

application/pdf

URL: https://www.lib.ku.ac.th/KUthesis/2561/nattapong-kur-all.pdf

CallNumber: TK7882.S65 .N38

eng

DegreeName: Master of Engineering

Level: Master's Degree

Descipline: Information and Communication Technology for Embedded System

Grantor: Kasetsart University

©copyrights Kasetsart University

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	nattapong-kur-all.pdf	91.81 MB

ใช้เวลา

0.040064 วินาที