The Development of a Prediction Model for the Acoustic Quality of Fans

Fan sounds are a typical part of environmental noise that humans hear in different everyday situations and they can play a major role in the appreciation of a certain product. Especially loud and unpleasant sounds can be detrimental in certain applications. Although it is well known that many technical acoustic measures like a-weighted sound pressure levels have limitations in depicting the perception of sound, they are still widely used because they are easily communicated and commonly used by engineers. However, using only a level-based measure as a sole indicator can be misleading in an optimization process especially for sounds differing in spectral content or temporal signatures. In order to enable a successful development of more pleasant fan sounds, it is therefore necessary to understand and characterize the perceptually relevant aspects of fan sounds and their impact on the evaluation of the sound in a certain context. A model for the prediction of the acoustic quality of fans is currently under development in a project funded by the German research association for air and drying technologies (FLT e.V.). Based on extensive listening tests and a factor analysis, the most important perceived sound characteristics could be identified for a broad variety of different fan sounds. Utilizing the output of a standardized loudness model, two indicators were developed that characterize the most important spectral aspects and distinguish three major groups of fan sounds. The pleasantness of the fan test sounds was assessed with an indirect listening test method adjusting the level of a fan sound to make it equally preferred as a reference sound. The level differences between test and reference sound can then be interpreted as a level penalty or level adjustment for the fan sound. The developed prediction model links the perceived sound characteristics described by the indicators with the fan sound evaluations from the listening tests measured as level differences. Using the model, differences in sound character can be translated to a dB value, which might be more commonly understood and more easily interpreted than a rating on a categorical scale. At present, the model is extended to cover a range of absolute sound pressure levels between 45 and 75 dB(A) and to characterize the impact of tonal components on the (un-)pleasantness assessment of the fan sound more precisely.