Loudness predictions using a physiologically based auditory model.
Item
-
Title
-
Loudness predictions using a physiologically based auditory model.
-
Identifier
-
AAI9224825
-
identifier
-
9224825
-
Creator
-
Jeng, Patricia Sheau-Feng.
-
Contributor
-
Adviser: Harry Levitt
-
Date
-
1992
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Health Sciences, Audiology | Physics, Acoustics | Psychology, Experimental | Psychology, Physiological
-
Abstract
-
Objectives of this dissertation were to predict loudness using a physiologically-based auditory model and to evaluate the model using loudness data. Loudness was studied because of its bearings in industrial and clinical applications and was used to evaluate the auditory model because of its dependence on harmonious functioning of many physiological mechanisms. The auditory model studied was the Ensemble Interval Histogram (EIH), which consisted of modules approximating functions of physiological units fundamental to audition. These units included middle ear, basilar membrane, inner hair cell, auditory nerves, and central auditory system. The loudness model (EIH-LM), based on the EIH model, was used to compute loudness values based on the assumption that loudness was proportional to the number of auditory neurons excited in the brain. Loudness data used for predictions included loudness functions (in sones), loudness level functions (in phons), and iso-loudness contours. For pure tones, published loudness data from Fletcher and Munson (1933), Hellman and Zwislocki (1976), and Scharf (1978) were used. For complex tonal stimuli, loudness data for four-tone complexes obtained by Scharf (1959) were used. In addition, an experiment was performed to obtain loudness level data for speech, narrow-band noise, and square wave stimuli. For each stimulus type, loudness levels were measured for the test stimulus in quiet, partially masked by noise, and for the total loudness level of stimulus plus noise. Data was obtained for two types of reference stimuli and two methods of adjustment. Experimental results showed that (1) variations within a subject were smaller than those between subjects; (2) loudness levels obtained with different reference stimuli did not differ significantly; (3) a regression effect associated with two adjustment methods was observed; (4) loudness level functions for the square wave differed significantly from those for speech and narrow-band noise. Various model parameters were manipulated to examine the effect of these parameters on loudness predictions. The results showed that the accuracy of loudness predictions for both pure tones and complex stimuli depended on (1) the frequency range covered by cochlear filters; (2) low-frequency slopes of low-frequency filters; (3) the number of neurons activated within each cochlear channel; (4) neuron firing rate; and (5) integration time.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
