Abstract
A soundscape assessment method that is suitable for the automatic
categorization of binaurally recorded sound in urban public places
is presented. Soundscape categories are established as a result of
an automatic clustering algorithm based on multi-parameter analysis
by 13 acoustical parameters used as similarity measures, on a large
set of sound recordings. One of the main advantages of the followed
approach allows to take into account an optimized set of parameters
that are judged relevant and necessary for an appropriate description
of the sampled acoustical scenarios. The Euclidian distance based
clustering of the 370 recordings of typical situations based on these
parameters, allows to categorize each binaurally recorded sound sample
into one of 20 proposed clusters (soundscape categories). The common
features among members within each cluster allow to identify “how
the acoustical scenario of the members sounds like”. The hybrid use
of an optimized set of standard acoustical quantities, such as sound
pressure level, together with well known psychoacoustical parameters
that directly relate to human perception of sound, makes the propose
method very robust.
Users
Please
log in to take part in the discussion (add own reviews or comments).