Comparison of Classifiers for Speech Emotion Recognition (SER) with Discriminative Spectral Features

Palo, Hemanta Kumar; Behera, Debasis; Rout, Bikash Chandra

doi:10.1007/978-981-15-2774-6_10

Cited by 9 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The reduced feature vector is classified by a Gaussian elliptical basis function neural network classifier. Palo et al proposed an SER system in wavelet domain based on Mel-frequency coefficients [10]. Both static and dynamic elements of the coefficients are combined for an SER system.…”

Section: Introductionmentioning

confidence: 99%

Emotion classification from speech signal based on empirical mode decomposition and non-linear features

Krishnan

Raj

Rajangam

2021

Complex Intell. Syst.

View full text Add to dashboard Cite

Emotion recognition system from speech signal is a widely researched topic in the design of the Human–Computer Interface (HCI) models, since it provides insights into the mental states of human beings. Often, it is required to identify the emotional condition of the humans as cognitive feedback in the HCI. In this paper, an attempt to recognize seven emotional states from speech signals, known as sad, angry, disgust, happy, surprise, pleasant, and neutral sentiment, is investigated. The proposed method employs a non-linear signal quantifying method based on randomness measure, known as the entropy feature, for the detection of emotions. Initially, the speech signals are decomposed into Intrinsic Mode Function (IMF), where the IMF signals are divided into dominant frequency bands such as the high frequency, mid-frequency , and base frequency. The entropy measures are computed directly from the high-frequency band in the IMF domain. However, for the mid- and base-band frequencies, the IMFs are averaged and their entropy measures are computed. A feature vector is formed from the computed entropy measures incorporating the randomness feature for all the emotional signals. Then, the feature vector is used to train a few state-of-the-art classifiers, such as Linear Discriminant Analysis (LDA), Naïve Bayes, K-Nearest Neighbor, Support Vector Machine, Random Forest, and Gradient Boosting Machine. A tenfold cross-validation, performed on a publicly available Toronto Emotional Speech dataset, illustrates that the LDA classifier presents a peak balanced accuracy of 93.3%, F1 score of 87.9%, and an area under the curve value of 0.995 in the recognition of emotions from speech signals of native English speakers.

show abstract

Section: Introductionmentioning

confidence: 99%

Emotion classification from speech signal based on empirical mode decomposition and non-linear features

Krishnan

Raj

Rajangam

2021

Complex Intell. Syst.

View full text Add to dashboard Cite

show abstract

“…Studies have shown that audio features such as the long-term structure of the audio spectrum, 35 Mel Frequency Cepstral Coefficient (MFCC), 36,37 its derivatives and transformations, 11,16,17,38,39 and Perceptual Linear Predictive, 36,37 are useful for gender classification and emotion recognition. Acoustic and prosodic features of speech 14,15,25 are other techniques that have been used for this classification and recognition.…”

Section: Introductionmentioning

confidence: 99%

“…Acoustic and prosodic features of speech 14,15,25 are other techniques that have been used for this classification and recognition. Also, Gaussian Mixture Model (GMM), 14,38 SVM, 9,37 Artificial Neural Network, 40 and Deep Neural Network are among the "state-of-the-art" tools, which are used for feature extraction 41 and gender classification. [4][5][6]42,43 In another study, Livieris et al 44 used an ensemble semi-supervised self-labeled algorithm, known as iCST-voting, in solving the GR problem.…”

Section: Introductionmentioning

confidence: 99%

Laughter signature, a new approach to gender recognition

Folorunso

Popoola

Asaolu

2020

Engineering Reports

View full text Add to dashboard Cite

Gender Recognition (GR) is the process of identifying gender difference by extracting and evaluating features from data existing as images, video, audio, text, and other signals. The task of GR has been achieved using face, keystroke, gait, and speech features. "Laughter" with its intrinsic usefulness for this task is yet to be explored. Laughter is an important paralinguistic feature in human communication. It can change the meaning of speech when triggered by some form of arousal or amusement. Although, laughter can be "acted," but the human natural laughter is a spontaneous reflex response, which reasonably embeds some characteristic peculiarities of the individuals. The human brain has capacity to make this distinction. In this study, spontaneous laughter bouts of 123 volunteers (41 females and 82 males) were recorded. The Dynamic-Average of the Mel frequency Cepstral Coefficient (DA-MFCC) were generated and trained using two conventional and effective machine learning algorithms that have been employed in gender identification. These algorithms are Gaussian Mixture Model (GMM) and Support Vector Machine (SVM) classifiers. Overall accuracies of 87.65% and 86.91% were obtained with GMM and SVM respectively. Therefore, indicating the possibility of using laughter characteristics as signatures for distinguishing between male and female genders. For both classifiers, the use of DA-MFCC reasonably reduced training time. Some of the potential areas of applications of GR include security, health care, marketing, human machine interaction toward enhanced emotion recognition, automatic speaker recognition and forensics.

show abstract

Internet of Things: A Key to Unfasten Mundane Repetitive Tasks

Palo

Sahoo

2021

Integration of Cloud Computing With Internet of Things

View full text Add to dashboard Cite

Comparison of Classifiers for Speech Emotion Recognition (SER) with Discriminative Spectral Features

Cited by 9 publications

References 7 publications

Emotion classification from speech signal based on empirical mode decomposition and non-linear features

Emotion classification from speech signal based on empirical mode decomposition and non-linear features

Laughter signature, a new approach to gender recognition

Internet of Things: A Key to Unfasten Mundane Repetitive Tasks

Contact Info

Product

Resources

About