X. Liu, R. Seiilova-Olson
Tenvos Inc., California, United States
Keywords: Voice biomarkers, Large acoustic models, Impairment detection, Fatigue Monitoring
Fatigue from erratic and insufficient sleep remains among the most consequential and persistently unsolved threats to the US military. The 2021 Office of the Secretary of Defense Report to Congress on Sleep Deprivation and Warfighter Readiness designates sleep deprivation as potentially the most important biological factor determining service-member health and combat effectiveness, and documents that 64.3% of active-duty service members fail to meet the DoW minimum 7-hour sleep standard. A 2024 GAO review catalogued 130 fragmented DoW fatigue research projects since 2017, none of which produced a unified, field-deployable, objective monitoring solution. Voice communication is used extensively in the military but it has not been used for inferring warfighter state in a scalable and objective manner. Common states such as fatigue and hypoxia impact human speech in a manner that can be detected using machine learning and advanced signal processing, yet they haven’t been turned into a product that can support the pilots. Integration of such technology into existing communication systems can enable continuous, non-invasive and objective monitoring of pilot state during training and flights. Private sector proof-of-concept are ongoing with a transportation fleet and a large petrochemical manufacturing plant to support fatigue risk reduction.