Emergency services worldwide are deploying AI voice systems in 911 dispatch centers based on efficiency metrics: speed, accuracy, cost reduction. These systems ignore the only variable that determines survival, whether the responder’s voice regulates or dysregulates a caller’s nervous system. This paper introduces a methodological framework for measuring physiological responses to vocal tone. The research proposed here should have been conducted before AI deployment began. Drawing on autonomic nervous system research and multiple theoretical frameworks including polyvagal theory, classical autonomic models, and limbic system neuroscience, we present the first framework for measuring vocal trust in crisis contexts: a composite measure of physiological, behavioral, and outcome based indicators. We propose pilot studies measuring caller heart rate, biometric indicators, vocal stress markers, and survival outcomes. Without this validation, emergency services are conducting a global experiment on millions of people during the most vulnerable moments of their lives.

Updated license to facilitate broader research collaboration.