Yasuhiro Ono, Joshua Lifton, Mark Feldmeier, Joseph A. Paradiso

Abstract

In this paper, we introduce distributed acoustic conversation shielding, a novel application of a transducer (sensors and speakers) network. This application protects the privacy of spontaneous conversations in a workplace by masking the participants’ voices with sound from distributed loudspeakers that adapt to the dynamic location of the conversation vs. that of potential eavesdroppers. We demonstrate how the speakers collaborate with various sensors to produce masking sounds that satisfy the requirements of this application. An index of intelligibility, SNR (Signal-toNoise Ratio) was used to evaluate the performance of our system. We suggest how the measured SNR can be used to adaptively servo the volume of the masking sounds.