In the historic center of Alghero, a seaside town in Sardinia-Italy, a soundwalk with 18 participants was performed. Ten sites distributed along the soundwalk path have been selected to be representative of different environments, namely in pedestrian areas, in an urban garden, along the seafront and so forth. Binaural recordings were carried out with simultaneous subjective appraisals of the sonic environment and of other features, like perceived landscape quality and its influence on soundscape ratings. The subjective data were collected by a questionnaire filled in by the subjects at each site. Acoustic parameters have been determined from binaural recordings and together with the subjective data have been analyzed by statistical procedures of feature extraction and classification, as well as to develop a model to predict classification membership. The unsupervised algorithm of hierarchical clustering was applied. Among the solutions the one that groups the data into three categories looked the most appropriate considering the characteristics of the sites. In order to develop a model for predicting the cluster membership, multinomial logistic regression has been performed by the “caret” package available in the “R” software using k-fold cross validation (k=10) and 5 repetitions. The available data have been split into two sets, one used for training the model and the other used for testing it. The results in terms of classification performance indices are rather promising.

SOUNDSCAPE CHARACTERIZATION AND CLASSIFICATION: A CASE STUDY

MASULLO, Massimiliano
2017

Abstract

In the historic center of Alghero, a seaside town in Sardinia-Italy, a soundwalk with 18 participants was performed. Ten sites distributed along the soundwalk path have been selected to be representative of different environments, namely in pedestrian areas, in an urban garden, along the seafront and so forth. Binaural recordings were carried out with simultaneous subjective appraisals of the sonic environment and of other features, like perceived landscape quality and its influence on soundscape ratings. The subjective data were collected by a questionnaire filled in by the subjects at each site. Acoustic parameters have been determined from binaural recordings and together with the subjective data have been analyzed by statistical procedures of feature extraction and classification, as well as to develop a model to predict classification membership. The unsupervised algorithm of hierarchical clustering was applied. Among the solutions the one that groups the data into three categories looked the most appropriate considering the characteristics of the sites. In order to develop a model for predicting the cluster membership, multinomial logistic regression has been performed by the “caret” package available in the “R” software using k-fold cross validation (k=10) and 5 repetitions. The available data have been split into two sets, one used for training the model and the other used for testing it. The results in terms of classification performance indices are rather promising.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11591/372428
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 3
  • ???jsp.display-item.citation.isi??? ND
social impact