Modeling subjective evaluation of soundscape quality in urban open spaces: An artificial neural network approach

Abstract

This research aims to explore the feasibility of using computer-based models to predict the soundscape quality evaluation of potential users in urban open spaces at the design stage. With the data from large scale field surveys in 19 urban open spaces across Europe and China, the importance of various physical, behavioral, social, demographical, and psychological factors for the soundscape evaluation has been statistically analyzed. Artificial neural network (ANN) models have then been explored at three levels. It has been shown that for both subjective sound level and acoustic comfort evaluation, a general model for all the case study sites is less feasible due to the complex physical and social environments in urban open spaces; models based on individual case study sites perform well but the application range is limited; and specific models for certain types of location/function would be reliable and practical. The performance of acoustic comfort models is considerably better than that of sound level models. Based on the ANN models, soundscape quality maps can be produced and this has been demonstrated with an example.

Effect of sound-related activities on human behaviours and acoustic comfort in urban open spaces

“…To study the influence of the activities on the evaluation of acoustic comfort, some questionnaire surveys were also conducted at the case site (Yu and Kang 2009). The questionnaire survey was conducted immediately after behaviour measurement, and every questionnaire survey was generally done by the interviewer in 3-5 minutes (Litwin 1995).…”

Ten questions on the soundscapes of the built environment

“…The hypothesis for developing such maps is that based on the soundscape information on certain locations in an area, the soundscape of the whole area could be predicted with spatial interpolation analysis method in GIS platforms. On the other hand, under a planning and design perspective, researchers have been exploring the possibility of using computer-based models, like Artificial Neural Networks (ANNs), to predict the soundscape quality evaluation of potential users in urban open spaces at the design stage (Yu & Kang, 2009). Such an approach requires large amount of data for the models to be likely to generalise and further research in this direction would be desirable.…”

Effects of typical dining styles on conversation behaviours and acoustic perception in restaurants in China

“…After the measurement of crowd density and sound pressure level, some diners were immediately extracted and invited to take a questionnaire survey (questions and scales are in Table 2). The questions covered diners' social characteristics, and acoustic comfort [57]. The interviewers were instructed to explain questions and ensure that interviewees understood them.…”

“…The interviewees, who were diners chosen randomly from the case sites, were asked to assess the acoustic comfort of the restaurant. On acoustic comfort, interviewees answered on the following five-point Likert-type scale: 1, very uncomfortable; 2, uncomfortable; 3, neither comfortable nor uncomfortable; 4, comfortable; 5, very comfortable [57]. Before the formal investigation, the validity and reliability of the questionnaire were tested [58][59][60].…”