Voice localization using nearby wall reflections

A Scalable Solution for Signaling Face Touches to Reduce the Spread of Surface-based Pathogens

“…Second, due to the relatively close distance between the hand and the sonar sensors (earset and mic), one cannot approximate the hand as a single point reflector (or scatterer); this is because acoustic signals may reflect off multiple sections of the user's hand. Addressing this problem is particularly challenging in the context of Saving Face since it cannot employ arrays for imaging [38] and must rely entirely on a single sensor (earset-mic pair). The machine learning algorithm was thus built to augment the system's capacity to detect distortion patterns in the ultrasound signal spectrogram that are unique to a face-touching gesture.…”

We Hear Your PACE

“…Recently appeared passive localization lifts the "burden" off users by passively tracking disturbances caused by user presence or motion. Such disturbances can be sensed by monitoring Wi-Fi communications [1,19,20,39] or ambient fields/signals [17,34,45]. However, there are small-scale indoor applications that existing solutions fail to handle; we illustrate such applications by two scenarios.…”

“…The active localization approach (e.g., [4,23,25,54,62]) is largely infeasible for both cases, as the users may feel cumbersome to carry a device or be unwilling to get tracked, so a passive approach is apparently preferred. Unfortunately, passive localization systems based on Wi-Fi and PIR [19,20,34,39] may suffer severe co-channel interference (especially the communication function of Wi-Fi devices), while Platypus [17] and VoLoc/Symphony [45,53] require either a heavy sensing infrastructure or user voice to perform localization. Therefore, the open question is: for small-scale indoor scenarios under a short coverage radius, can we passively track multiple users without heavy infrastructural support and user involvement?…”

“…However, implementing PACE faces two practical challenges. On one hand, indoor environments normally incur a heavy multi-path effect and also have a strong acoustic noise background; these have made pure angle of arrival (AoA) enabled solutions (e.g., [45]) insufficient to locate footsteps. On the other hand, while it is intuitive that human hearing can clearly identify acquainted footsteps, this identification is not readily achievable via acoustic sensing.…”

“…In this section, we present extensive evaluation studies on our PACE prototype. We refrain from system-level comparisons with other up-to-date passive indoor localization proposals (e.g., [38,45]), because their application scenarios are very different as explained in Section 1.…”