Privacy-Aware Acoustic Assessments of Everyday Life

There is not much data available on the acoustical properties of our everyday environments and our subjective impressions and communication abilities in those environments. In this paper we present a smartphone-based system that is capable of measuring the most important features of the surrounding acoustics. However, in contrast to simple audio recordings and subsequent analysis the proposed solution will respect the privacy of all communication partners and bystanders by extracting a feature set tested for privacy compliance only. Nonetheless, a larger set of necessary features for acoustical environment classification can be computed in the necessary accuracy afterwards in an off-line process. For a given feature set the statistical analysis shows comparable results in the extracted data, when either the original audio signals or the new privacy-aware extraction methods are used.