Perceiving the sources of environmental sounds with a varying number of spectral channels.
Item
-
Title
-
Perceiving the sources of environmental sounds with a varying number of spectral channels.
-
Identifier
-
AAI3115289
-
identifier
-
3115289
-
Creator
-
Shafiro, Valeriy.
-
Contributor
-
Adviser: Winifred Strange
-
Date
-
2004
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Psychology, Experimental | Health Sciences, Audiology | Psychology, Cognitive
-
Abstract
-
Nonspeech environmental sounds provide the listener with valuable information about sound-producing objects and events in the immediate environment. Avoiding a collision with an unseen car, responding to a ringing doorbell, or enjoying a bird song are only a few examples of the potential benefits of accurate object and event identification. However, relatively little is known about the cognitive factors and acoustic parameters that influence the perception of the sources of environmental sounds. A major difficulty associated with environmental sound research is the great variety in the types of sound sources in the environment, and the lack of a clear taxonomy of environmental sounds. In an attempt to provide an empirical basis for environmental sound classification and explore the acoustic parameters important in environmental sound perception, the present research investigated the effects of systematically varying the spectral resolution of familiar environmental sounds on listeners' ability to identify the corresponding sound sources.;Normally hearing adult listeners were asked to identify the sources of 60 familiar environmental sounds processed through a vocoder simulation of a cochlear implant with varying numbers of frequency channels. Using a Latin square design, listeners heard 10 different sounds in each of six channel conditions (2, 4, 8, 16, 24, 32) followed by all 60 undistorted sounds. For each sound, listeners selected one of 60 response options that best described the source of the sound. Results indicate that identification performance continuously improved with an increasing number of frequency channels and reached 81% correct with 32 channels. However, for some sounds, identification accuracy declined when the number of channels was 16 or higher due, perhaps, to the spectral asynchrony introduced by unequal filter group delays across channels. Thus, overall identification accuracy did not substantially improve beyond 16 channels. In general, broadband temporally patterned sounds (e.g., 'helicopter', 'clapping', 'typing') tended to require fewer channels to be identified correctly than sounds with time-varying narrow band resonances (e.g., 'baby crying', 'doorbell', 'rooster'). These findings, compared with those of other investigators, suggest that, without training, the accurate perception of environmental sounds may require more spectral channels than that of speech sounds.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
