Voice In Ear

Abstract

With the rapid growth of wearable computing and increasing demand for mobile authentication scenarios, voiceprint-based authentication has become one of the prevalent technologies and has already presented tremendous potentials to the public. However, it is vulnerable to voice spoofing attacks (e.g., replay attacks and synthetic voice attacks). To address this threat, we propose a new biometric authentication approach, named EarPrint, which aims to extend voiceprint and build a hidden and secure user authentication scheme on earphones. EarPrint builds on the speaking-induced body sound transmission from the throat to the ear canal, i.e., different users will have different body sound conduction patterns on both sides of ears. As the first exploratory study, extensive experiments on 23 subjects show the EarPrint is robust against ambient noises and body motions. EarPrint achieves an Equal Error Rate (EER) of 3.64% with 75 seconds enrollment data. We also evaluate the resilience of EarPrint against replay attacks. A major contribution of EarPrint is that it leverages two-level uniqueness, including the body sound conduction from the throat to the ear canal and the body asymmetry between the left and the right ears, taking advantage of earphones' paring form-factor. Compared with other mobile and wearable biometric modalities, EarPrint is a low-cost, accurate, and secure authentication solution for earphone users.

Beyond Accuracy: A Critical Review of Fairness in Machine Learning for Mobile and Wearable Computing

“…[152]. On the contrary, in speech-based human identification scenarios, false positive outcomes are critical in preventing unauthorized access, as "existing voiceprint-based authentication often suffers from various voice spoofing attacks" [40].…”

“…Device-related components include device type, sampling rate, and operating system, as well as device placement and orientation in activity and gaze tracking [53,71], vital sign monitoring and physiological sensing [78,136], speech recognition via built-in sensors and speech synthesis [73,125], and user behavior sensing [60]. Environmental components include ambient noise, light, and temperature that might affect data quality of acoustic [40,71,136] or video [44,78,133,139] signals, respectively, or random passers-by that might affect model performance on the individual for human identification [145]. Regarding experimental setup, few included papers studied the effect of equipment placement (i.e., distance, angle) and characteristics (i.e., range) on the model's robustness in activity sensing [71] and vital sign monitoring using acoustic signals [43,136].…”

Pattern unlocking guided multi‐modal continuous authentication for smartphone with multi‐branch context‐aware representation learning and auto encoder

“…Liu et al 22 leverages low‐cost vibration signal on a solid input interface, that is, the smartphone screen, to discriminate user's finger input. Researchers of References 23 and 24 attempt to collect acoustic data through a microphone on the smartphone to portray the biometric features of the user. Successfully avoided the introduction of dedicated sensors though, these solutions have not fully discussed on over‐time changes in users' voiceprint features, and are prone to be attacked 25 …”

Decoding silent speech commands from articulatory movements through soft magnetic skin and machine learning

“…Contactless approaches are mainly explored through camera-based visual signals, [9][10][11][12][13][14][15] ultrasound signals. [16][17][18][19][20][21][22][23] Camerabased visual solutions require external video tracking devices, and users must remain within the camera's line of sight. Despite efforts to develop compact shoulder-mounted devices 9 to enhance portability, visual solutions still face challenges in terms of lighting conditions and angles between users and cameras, thereby limiting their practicality.…”

“…As a more portable and user-friendly ultrasound-based solution, the speaker of the cell phone was used to emit ultrasound signals, and the microphone was employed to capture reflected signals from the lips. 16,[18][19][20][21][22] This method is not hands-free and is susceptible to multipath interferences caused by bodily movements and surrounding objects.…”