O.V. Melnik1, S.I. Babaev2, M.N. Saraev3

1–3 FSBEI HE «Ryazan State Radio Engineering University named after V.F. Utkin» (Ryazan, Russia)

1 omela111@yandex.ru, 2 babaev.s.i@gmail.com, 3 mixailr@mail.ru

Human emotional states manifest in speech characteristics, particularly under conditions of significant stress or psychological pressure. Optimizing the set of acoustic features is crucial for reducing computational complexity in automated stress assessment algorithms and preventing overfitting. A limited yet informative feature set enhances classifier generalizability while maintaining result interpretability. This study proposes a methodology for selecting speech acoustic features suitable for automated emotional stress evaluation.

To develop a feature selection methodology incorporating stepwise analysis, statistical evaluation, and visualization techniques for feature space optimization.

A comprehensive analysis of speech acoustic features was conducted to optimize the feature set for automatic stress detection. The selected feature subset demonstrated optimal separability between "normal" and "stress" speech segments, recommending its use in automated classification systems.

The proposed methodology enables automated detection of emotional stress in speech, with potential applications in psychology, clinical diagnostics, and human state monitoring systems.

Melnik O.V., Babaev S.I., Saraev M.N. A feature selection methodology for speech-based automated emotional stress detection // Biomedicine Radioengineering. 2026. V. 29. № 5. Р. 73-77. DOI: https://doi.org/10.18127/j15604136-202605-11

1. Kuchibhotla S., Vankayalapati H.D., Anne K.R. An optimal two stage feature selection for speech emotion recognition using acoustic features. Int. J. Speech Technol. 2016. V. 19. P. 657–667.
2. Pravena D., Govind D. Significance of incorporating excitation source parameters for improved emotion recognition from speech and electroglottographic signals. Int. J. Speech Technol. 2017. V. 20. P. 787–797.
3. Buanzur T., Zellers M., Namyalo S., Witzlack-Makarevich A. A First Investigation of the Timing of Turn-taking in Ruuli. Proceedings of the 19th Annual Conference of the International Speech Communication Association (INTERSPEECH, 2018). P. 621–625.
4. Melnik O.V., Babaev S.I., Saraev M.N. Analysis of Acoustic Characteristics of Speech for Assessing Emotional Stress: Methods and Experimental Studies. Proceedings of the 2025 International Conference on Systems and Technologies of the Digital HealthCare (STDH – 2025) June 2-6. 2025. P. 84–89.
5. Yanchi Liu, Zhongmou Li, Hui Xiong et al. Understanding of Internal Clustering Validation Measures. 2010 IEEE International Conference on Data Mining. Sydney, NSW, Australia. 2010. P. 911–916.