Advanced Deep Learning Techniques with Attention Mechanisms for Acoustic Emotion Classification

Speech Emotion Recognition (SER) is the most prominent and emerging field in the digital signal processing area. Although plenty of research work has been carried out in this area, emotion recognition in a real-time environment is very challenging. This is because of the increased number of paramete...

Full description

Saved in:
Bibliographic Details
Main Authors: Jothimani, S., Sangeethaa, S. N., Premalatha, K.
Format: Conference Proceeding
Language:English
Subjects:
Acoustics
Attention mechanism
CNN
Deep learning
Digital signal processing
Emotion recognition
Feature extraction
LSTM
MFCC
Real-time systems
speech databases
Speech recognition
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Speech Emotion Recognition (SER) is the most prominent and emerging field in the digital signal processing area. Although plenty of research work has been carried out in this area, emotion recognition in a real-time environment is very challenging. This is because of the increased number of parameters considered for emotion recognition, like the voice, tone, pitch, and quality of the audio signal. The previous research papers do not cover all the deep learning techniques and the feature extraction techniques for speech emotion recognition. In this paper, the techniques used to analyze the emotion along with the recording environments are reviewed. The state-of-the-art deep learning techniques are reviewed and the dataset and feature extraction techniques used for this analysis are also considered in this paper.
ISSN:2767-7788
DOI:10.1109/ICICT54344.2022.9850908