Your voice may indicate your stress level

Chronic stress is expensive both medically and economically: it is associated with depression, which affects 5% of the world’s population, reduces motivation and creativity, and affects labor productivity. The search for simple and non-invasive ways to measure stress has been going on for a long time, and human speech has become one of the candidates for the role of such a tool.

When the body is in a state of tension, the sympathetic nervous system increases the tone of the muscles involved in speech: the muscles of the larynx, respiratory muscles, tongue, lips and jaws. The voice changes in terms of pitch, intonation, volume, and timbre, and the profiles of these changes differ under mental and physical stress. Artificial intelligence algorithms have already learned to recognize stress based on acoustic parameters of speech with an accuracy of 70% to 90%, depending on the methodology, and the authors of the new study have achieved an accuracy of about 86% in previous research by comparing recordings before and after therapeutic intervention.

This time, the researchers decided to test whether a single 20-minute relaxation session would be sufficient to induce a measurable shift in prosody. Thirty participants were divided into three groups: the first group underwent a guided mindfulness meditation session, the second group received vibroacoustic stimulation, in which an audible sound is accompanied by a synchronous tactile vibration, and the third group simply waited in silence. Before and after the session, each participant read the same text aloud, and the researchers compared the acoustic parameters of the two recordings.
The results were divided into three dimensions: intonation, voice quality (the degree of tension or relaxation), and volume.

The most unexpected finding was that the most noticeable changes occurred in the control group, rather than in the intervention groups. After twenty minutes of silence, the participants in the control group read the text in a lower tone, with a wider range of pitch, but their voices were also more tense, and their volume decreased by 4-5 decibels. The authors suggest that the forced inactivity in silence may have caused stress or mental lethargy, which affected their speech.

In contrast, the participants in the meditation and vibroacoustic stimulation groups showed little change in their speech patterns after the session. The only parameter that shifted in the expected direction for both groups was the quality of their voices, which became more breathy after the intervention, a characteristic commonly associated with a relaxed physiological state, although this alone does not prove a reduction in stress.

The authors acknowledge that the hypothesis was only partially supported, and only for one of the several predicted prosodic parameters. According to their assessment, twenty minutes of relaxation is too short a time to cause stable and reliably detectable changes in speech. The team describes all interpretations as preliminary and exploratory, rather than as objectively confirmed conclusions.

Of course, the experiment has significant limitations: a small sample size of 30 participants, the lack of experimental verification of the proposed mechanisms, and the relatively low statistical threshold typical of pilot studies. To transform speech prosody from a promising idea into a practical tool for stress monitoring, large-scale longitudinal studies with multiple factors and comparisons of different types of interventions will be required.