MFCCs and TEO-MFCCs for Stress Detection on Women Gender through Deep Learning Analysis

Men and women describe differing physical and emotional responses to stress; women reported experiencing it more than men with 11.7% higher. This issue has been affecting women in different ways than men due to biological and social factors (e.g., differences in hormone processes between both genders and dual responsibilities in the workplace as well as at home). This crucial issue raises many concerns about women’s mental health, and prolonged stress, such as heart problems, sleep problems, and others, will ideally impact them. Early stress detection is a crucial strategy to overcome the said problems since mental health issues always begin with stress problems. Therefore, in this paper, the MFCCs and TEOMFCCs for stress detection in the women’s gender through deep learning are presented. The stress classification had been made by utilizing the speech features, which are Mel Frequency Cepstral Coefficients (MFCCs) and Teager Energy Operator-Mel Frequency Cepstral Coefficients (TEO-MFCCs), with the help of Deep Learning technology, which is Convolutional Neural Networks (CNNs). The Toronto Emotional Speech Set (TESS) has been selected for this study since it consists of women’s speech data. The outcome shows that MFCCs provide better accuracy in predicting women’s stress, with a 98% score outperformed another study using the same dataset.