Speech Emotion Recognition Using Energies in six bands and Multilayer Perceptron on RAVDESS Dataset

Emotional information has recently received in-creased attention in studies on speech signals. Selecting the ideal speech feature representation for speech emotion recognition (SER) is the most difficult task. According to the statistical study, the functions of each speech characteristic vary in va...

Full description

Saved in:
Bibliographic Details
Main Authors: Agrima, Abdellah, Barakat, Aziza, Mounir, Ilham, Farchi, Abdelmajid, ElMazouzi, Laila, Mounir, Badia
Format: Conference Proceeding
Language:English
Subjects:
Communications technology
Emotion recognition
Energy distribution
Feature extraction
Multilayer Perceptron
Multilayer perceptrons
Reliability
Signal processing
Speech emotion recognition
Speech recognition
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Emotional information has recently received in-creased attention in studies on speech signals. Selecting the ideal speech feature representation for speech emotion recognition (SER) is the most difficult task. According to the statistical study, the functions of each speech characteristic vary in various emotional states, suggesting that various features have varying capacities for emotion distinction. A feature extraction and classification technique to create a reliable recognition system is presented in this article. We also provide new features that are based on energy and energy distribution taken from the (RAVDESS) database. Based on the results of the experiments, it can be concluded that the suggested strategy, when compared to a direct (multi-class classification).
ISSN:2771-7402
DOI:10.1109/CommNet56067.2022.9993940