Person identification from walking sound on wooden floor

Paper i proceeding, 2018

Human activities in building structures may vary to a great extend and daily mobility may be the cause of noise and vibrations. We focused on walking sound to identify different individuals based on airborne sound recordings. Our aim was to provide a description of high-level acoustical features that corresponds to walking sound and person identification. We present two levels of abstraction. The first level builds upon principal component analysis and provides the main sound characteristics of walking activity. For the second level of abstraction we provide higher-level acoustical features that better describe person identification.