Perception Of Final Consonant "Voicing" In Phonated And Whispered Speech
Item
-
Title
-
Perception Of Final Consonant "Voicing" In Phonated And Whispered Speech
-
Identifier
-
d_2009_2013:958f3fcca0b7:11235
-
identifier
-
11641
-
Creator
-
Gilichinskaya, Yana D.,
-
Contributor
-
Winifred Strange | Glenis Long
-
Date
-
2012
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Audiology | Behavioral psychology | consonants | consonant-vowel duration ratio | percpetion | periodicity | production | whispered
-
Abstract
-
Our everyday verbal interaction does not always happen in the communication conditions optimal for the speaker or the listener. In order to increase intelligibility and to overcome interference of the communication medium, speakers may switch from a conversational rate of speaking to "clear" speech mode. In other situations, when fully phonated speech is not appropriate, speakers may deliberately depart from the optimal listening conditions to adopt yet another speaking mode---whispered speech. Despite acoustic differences between phonated and whispered speech, intelligibility of whispered speech remains relatively accurate. In perception of whispered consonants, the characteristic that is most susceptible to the absence of phonation is "voicing". At the same time, listeners identify "voicing" in whispered consonants far above chance. We hypothesized that perceptual discernment of whispered consonant "voicing" may be rooted in the maintenance and potential enhancement of time-based parameters that are known to contribute to consonant voicing: vowel duration, consonant duration and combined consonant-to-vowel duration-ratio. Acoustic parameters were compared between phonated and whispered conversational speech, and conversational and clear whispered speech. The results indicated that all time-based acoustic cues to "voicing" were maintained in whispered speech. In perception tests, overall consonant identification accuracy in conversational whispered speech was about 85% with errors on place or manner of articulation not exceeding 1%. The improvement in perception of clear whispered speech varied across speakers from -2% ... to 8%. The perceptual advantage afforded by two speakers was generally in the agreement with the predictions made on the basis of the production data. Logistic regression analysis indicated that after controlling for variability in speakers, consonant-pairs and vowels, consonant "voicedness" in both phonated and whispered speech could be accurately predicted by a single parameter---C/V duration-ratio. At the same time, correlation analysis between the magnitude of C/V duration-ratio contrast and consonant "voicing" intelligibility suggested that listeners did not fully utilize the distinctive information provided by the C/V duration-ratio. This study contributes to knowledge of factors responsible for intelligibility of whispered speech which may be valuable in designing speech processing algorithms for hearing-assistive devices and developing synthesis systems for speech reconstruction in voice-impaired patients.
-
Type
-
dissertation
-
Source
-
2009_2013.csv
-
degree
-
Ph.D.
-
Program
-
Speech and Hearing Sciences
