Biometric verification using oral acoustic reflections

Biometric verification methods, such as voice, face, iris and fingerprint verification, are commonly used in smartphones, security systems, and medical devices. However, each involves tradeoffs in cost, accuracy, latency and susceptibility to spoofing. We present and report laboratory testing of Oral Cavity Acoustic Verification (OCAV), a biometric verification system that transmits sound waves into the user's oral cavity, characterizes the acoustic response, and returns a verification decision by comparing the acoustic response to the individual's prerecorded acoustic signature. A total of 108 healthy volunteers aged 18-60 years were recruited, and a unique acoustic profile was created for each participant. Across three recording sessions conducted one week apart, 20,860 genuine trials and 18,880 impostor trials were collected to evaluate verification accuracy. Verification accuracy was 90.1% for a single trial and 95.2% when pooling 5 consecutive trials. Unlike voice biometrics that depend on speech production and are sensitive to acoustic noise and audio playback attacks, OCAV exploits an active acoustic response from the oral cavity that may be less vulnerable to spoofing attacks. In contrast to other audio-based systems such as ear-canal acoustic biometrics, OCAV is intended for devices that already involve oral cavity contact. The potential applications include inhaler-based health monitoring, breath-alcohol tests, and the enforcement of nicotine-use restrictions. These capabilities can help support public health and safety efforts; for example, by monitoring youth access to e-vaping or detecting drivers' alcohol consumption prior to ignition.