|
KMID : 1036820170220040818
|
|
Communication Sciences & Disorders
2017 Volume.22 No. 4 p.818 ~ p.826
|
|
|
A Comparison of Cepstral and Spectral Measures according to Measurement Position in a Reading Passage
|
|
Lee Seung-Jin
Lim Sung-Eun
Choi Hong-Shik
|
|
|
Abstract
|
|
|
Objectives: The purpose of the present study was to investigate whether cepstral and spectral measures differ according to measurement position in a reading passage, upon controlling for the overall severity of voice disorders. Furthermore, the diagnostic value of perceptual estimates in each position was explored.
Methods : Thirty-five patients with voice disorders and 35 normal controls were asked to read the standard passage ‘Ga-eul’. For each sample, sentence samples of 26-30 syllable length were trimmed in 3 positions (front, middle, rear) of the passage, measuring cepstral peak prominence (CPP), L/H spectral ratio, and standard deviations (SDs). Overall severity from the CAPE-V was measured perceptually. Upon controlling for severity, cepstral measures were compared between groups and positions. The association of acoustic measures with overall severity was explored by conducting a multiple regression analysis in each position. The diagnostic value of perceptual estimates derived from acoustic measures was explored using ROC curve analysis.
Results: Results indicated that CPP of the rear position was lower than that of the middle position, while SD of the L/H ratio of the rear position was higher than those of the front and middle positions. On the other hand, the relationship between perceptual and cepstral and spectral measures, along with the diagnostic value, was the highest for the front position.
Conclusion: It is recommended that the variability of the cepstral and spectral measures among positions be considered in research and clinical practice.
|
|
|
KEYWORD
|
|
|
Cepstral peak prominence (CPP), L/H spectral ratio, Acoustic analysis, Passage reading, Variability
|
|
|
FullTexts / Linksout information
|
|
|
|
|
|
Listed journal information
|
|
|
|