Abstract
Continuously-operating microphones on wearable computing devices (e.g., a smartwatch, augmented reality glasses, smart ring, etc.) may present challenges related to power consumption and user privacy, as well as difficulty in distinguishing a wearer's speech from ambient sounds. A disclosed technique can address these challenges through a bimodal sensing approach that concurrently analyzes air-conducted acoustic signals from a microphone and body-conducted kinetic signals from a motion sensor, such as an accelerometer or gyroscope. A processing system may perform a correlation analysis between features extracted from the acoustic data and the kinetic data to determine if a detected sound originates from the device wearer. The authentication of the wearer's speech can be used, for example, to selectively activate audio processing pipelines or other device functions. This technique may improve power efficiency by reducing processing of non-wearer sounds and can also address privacy considerations by potentially limiting the capture and analysis of ambient conversations.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Gore, Tyler; Rudolph, Aaron; and Ananth, Sharath, "Wearer Speech Authentication Using Correlated Acoustic and Body-Vibration Data", Technical Disclosure Commons, (March 25, 2026)
https://www.tdcommons.org/dpubs_series/9617