Solution study
Monday, December 09
11:45 AM - 12:15 PM
Live in Dearborn, Michigan
Less Details
The focus of our presentation “Voice As Health Biomarker: Sensor-less Approach For Automotive In-Cabin Physiological Sensing” lies in exploring the utilization of voice as a pivotal health biomarker, particularly in the automotive industry, through a sensor-less approach for in-cabin physiological sensing. By leveraging artificial intelligence and speech processing. The study delves into respiratory sensing embedded within speech, aiming to circumvent issues such as motion artifacts, subject variability, and privacy concerns linked with existing sensor-based methods. The proposed approach aims to provide a non-intrusive and on-demand, seamless method procedure for in-cabin sensing, prioritizing seamless integration and interoperability. The technology, from regular speech conversations, able to achieve approximate accuracy of +/-3 breaths per minute in estimating respiration rates for more than 85% of test subjects. This breakthrough holds significant advantages due to its simplicity in application and the convenience of utilizing everyday speech interactions as the basis for health monitoring within automotive environments. In a real-world scenario involving multiple speakers in an automotive cabin with varying signal-to-noise ratios and ambient noises, the algorithm proves its efficacy, performing sufficiently well in estimating the respiratory rate. The future work of the research involves extending speech-based sensing to other vitals including heart rate, PPG, etc., and further extend into in-cabin system personalization based on the speaker’s physiological parameters.