Detecting Public Speaking Stress via Real-Time Voice Analysis in Virtual Reality: A Review

Stress during public speaking is common and adversely affects performance and self-confidence of individuals in various professional contexts. Virtual reality (VR) has been used in psychology and human-computer interaction (HCI) to induce and measure barriers to good public speaking skills, such as anxiety and fear. We reviewed the most significant papers published on major indexed journals across the past 20 years. We outline how minimal research has been conducted to detect stress via auditory means in real time during public speaking. As a measurement criterion, the use of questionnaires and physiological parameters remains prevalent. Furthermore, to induce and provide feedback, past experiments relied on simulated audiences that are either scripted or controlled by outside human agents. In this context, we explore the current complexities, limitations and opportunities for novel systems that can provide more engaging and immersive experiences. Hence, we propose a conceptual framework for the development of a voice analysis-based stress-detection computational algorithmic model that can be integrated into a virtual reality simulation. The implementation of the proposed model would ultimately help users to gradually learn how to overcome their stress in real-time and improve their public speaking performance.