A Machine Learning-Based Approach for Non-Invasive Stress Analysis Using Speech Features for Enhanced Detection and Classification

Devika M., Shruthi R., Sahana K., Madhubala V.

2025

Abstract

Stress has its impact on our physical and mental health in many forms, and thus it becomes even more important to detect it at an early stage so that we can manage it better. Standard approaches to monitoring the stress response typically involve tracking a person’s physiological state or behavioural differences with the help of invasive sensors or elaborate data collection systems. As a more appealing non-intrusive option, this research paper proposes a machine learning model, to observe our speech patterns to monitor our stress levels on a real-time basis. By analysing features such as pitch, tone, frequency and the rate at which someone speaks, the system can gain a better understanding of a person’s emotional state. These pieces of information are collected by techniques like Mel Frequency Cepstral Coefficient (MFCC), Fast Fourier Transform (FFT), and Spectral Analysis. These extracted features are then used to train multiple machine learning classifiers such as Decision Trees and K-Nearest (K-NN) Neighbours. This enables you to classify stress levels into three specifics categories (Neutral, Mild Stress, High Stress). This analysis allows us to make sure the model is reliable and its performance is understood using standard metrics including but not limited to accuracy, precision, recall and F1-score. The idea is to make stress detection more achievable and operable every day.

Paper Citation

in Harvard Style

M. D., R. S., K. S. and V. M. (2025). A Machine Learning-Based Approach for Non-Invasive Stress Analysis Using Speech Features for Enhanced Detection and Classification. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 285-291. DOI: 10.5220/0013896800004919

in Bibtex Style

@conference{icrdicct`2525,
author={Devika M. and Shruthi R. and Sahana K. and Madhubala V.},
title={A Machine Learning-Based Approach for Non-Invasive Stress Analysis Using Speech Features for Enhanced Detection and Classification},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={285-291},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013896800004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - A Machine Learning-Based Approach for Non-Invasive Stress Analysis Using Speech Features for Enhanced Detection and Classification
SN - 978-989-758-777-1
AU - M. D.
AU - R. S.
AU - K. S.
AU - V. M.
PY - 2025
SP - 285
EP - 291
DO - 10.5220/0013896800004919
PB - SciTePress