Enhanced speech privacy in office spaces

Abstract

There is an ever-increasing demand for speech privacy in modern office spaces. People prefer their conversations to remain private while not being disturbed by speech of others. The work presented in this article focuses on reducing the intelligibility of the perceived speech by masking the original signal without a significant additional annoyance. The disruption of the information-carrying components of speech by the amplitude and temporal smearing is of a specific interest of this work. We suggest a method in which a modified self-adjusted masking signal is used to effectively counteract the drawbacks of a straightforward reverberation. The presented masking method disrupts key speech characteristics of the original signal. The intelligibility and annoyance levels of the resulting signal are subjectively evaluated. The optimized experimental parameters are reported. The presented method provides a significantly higher speech privacy and a lower perceived annoyance as compared to white noise. The method can be used in office environments with various Sound Transmission Class levels.