Azarias Reda, Saurabh Panjwani, Edward Cutrell
Tracking attendance is an important consideration for many developing world interventions. In many cases, these interventions are located in remote areas where its not always feasible to deploy expensive attendance tracking systems. In addition, since many existing systems focus on tracking participants (such as patients or students), rather than agents (such as teachers or health workers), they assume a trusted administrative staff on-site to record attendance. In this paper, we present the design of Hyke, a system for remote and cost effective attendance tracking in developing regions. Hyke combines voice-biometrics with accurate location tagging for tracking attendance in remote locations without the need for a trusted mediator on-site. Hyke was designed based on our observation of a currently deployed teacher attendance tracking system in rural Rajasthan, India. We have implemented some of the key components in Hyke, and discuss some of the security concerns in the system. The Hyke biometric stack for voice recognition is built atop several open source technologies, and provides a simple interface for non-expert users. Our evaluations with Indian speakers over telephone audio suggests the biometric stack is at par with the current state of the art. We believe this will be a useful tool for researchers who would like to incorporate voice technologies in their developing world projects.