Feature extraction and selection methods for emotion and stress recognition from natural speech / \c Yogesh Chinnakalai
This thesis aims to develop emotion/stress recognition system by using speech signals. This work mainly focuses on non-linear HOSA features, feature selection and machine learning algorithms that can represent the hidden and the minute information of emotions and stress present in speech signals ; t...
Saved in:
| Main Author: | |
|---|---|
| Corporate Author: | |
| Format: | Thesis Software eBook |
| Language: | English |
| Published: |
Perlis, Malaysia
School of Computer and Communication Engineering, Universiti Malaysia Perlis
2017
|
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This thesis aims to develop emotion/stress recognition system by using speech signals. This work mainly focuses on non-linear HOSA features, feature selection and machine learning algorithms that can represent the hidden and the minute information of emotions and stress present in speech signals ; to investigate and identify HOSA based feature set for efficient speech emotion recognition system ; to develop optimization Based Feature Selection Algorithms (BBPSO1 and BBPSO2) to select the suitable features ; to validate the selected features from the proposed feature selection algorithms in recognizing stress/emotions from speech signals. |
|---|---|
| Physical Description: | 1 CD-ROM 12 cm |
| Bibliography: | Includes bibliographical references. |