RTI uses cookies to offer you the best experience online. By clicking “accept” on this website, you opt in and you agree to the use of cookies. If you would like to know more about how RTI uses cookies and how to manage them please view our Privacy Policy here. You can “opt out” or change your mind by visiting: http://optout.aboutads.info/. Click “accept” to agree.
Multilevel annoyance modelling of short environmental sound recordings
Orga, F., Mitchell, A., Freixes, M., Aletta, F., Alsina-Pages, R. M., & Foraster, M. (2021). Multilevel annoyance modelling of short environmental sound recordings. Sustainability, 13(11), Article 5779. https://doi.org/10.3390/su13115779
The recent development and deployment of Wireless Acoustic Sensor Networks (WASN) present new ways to address urban acoustic challenges in a smart city context. A focus on improving quality of life forms the core of smart-city design paradigms and cannot be limited to simply measuring objective environmental factors, but should also consider the perceptual, psychological and health impacts on citizens. This study therefore makes use of short (1-2.7 s) recordings sourced from a WASN in Milan which were grouped into various environmental sound source types and given an annoyance rating via an online survey with N=100 participants. A multilevel psychoacoustic model was found to achieve an overall R-2=0.64 which incorporates Sharpness as a fixed effect regardless of the sound source type and Roughness, Impulsiveness and Tonality as random effects whose coefficients vary depending on the sound source. These results present a promising step toward implementing an on-sensor annoyance model which incorporates psychoacoustic features and sound source type, and is ultimately not dependent on sound level.