Use this URL to cite or link to this record in EThOS:
Title: Natural and urban sounds in soundscapes
Author: Yang, Ming
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
Among various sounds in the environment, natural sounds, such as water sounds and birdsongs, have proven to be highly preferred by humans, but the reasons for these preferences are not yet completely understood. This research study explores the differences between various natural and urban environmental sounds from the viewpoint of objective measures. Moreover, since numerous studies of soundscape perception and evaluation have revealed that besides the conventional parameters, e.g., A-weighted sound pressure level, additional parameters are necessary for soundscape measurement, in this study more possible parameters are explored. From alternative algorithms of the features proposed in literature for both perception of the auditory system and practical application in music and speech, the algorithms applicable for environmental sound are searched through comparison. The sound samples used in this study include the recordings of single sound source categories of water, wind, birdsongs, and urban sounds including street music, mechanical sounds and traffic noise. The samples are analysed with a number of objective parameters in three aspects, which include psychoacoustic parameters that have been recommended in previous soundscape researches, additional psychoacoustically related parameters that have previously mainly been applied in music perception, and 1/f noise dynamic that has been observed in music, speech, and soundscapes. Based on one-way analysis of variance, hierarchical cluster, and principal components analyses of the calculated results, a series of differences are shown among different sound types in terms of key parameters, which include fluctuation strength, pitch, loudness, and 1/f noise. Generally, both water and wind sounds have low fluctuation strength, pitch values, and pitch strengths; birdsongs have high fluctuation strength, pitch values, and pitch strength, low loudness, and exhibit generally 1/f behaviour of loudness in short and medium time intervals; and urban sounds have low pitch values, high loudness, and relatively wide ranges of other parameters. With the parameters, furthermore, the sound categories of recordings are automatically identified/classified using discriminant function analysis and artificial neural networks. With the artificial neural networks, which have better performance than the discriminant functions for the identification, based on all the psychoacoustic, music, and 1/f noise indices, the prediction accuracies are above about 99% for the three natural sound categories, i.e., of water, wind, birdsongs, and about 90% for the urban sound category.
Supervisor: Kang, Jian Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available