Abstract
A recently developed speech production model, in which speech segments are specified by relative acoustic events called resonance deflection patterns, was used to generate speech signals that were presented to listeners in a perceptual test. The purpose was to determine the effect of variations of the magnitude and polarity of the third resonance deflection on identification of the consonant in a V1CV2 disyllable while the deflections of the first and second resonances were held constant. Result showed that listeners' identification changed from /d/ to /ɡ/ when the polarity of the third resonance deflection switched from positive to negative.
Highlights
A recently developed speech production model, in which speech segments are specified by relative acoustic events called resonance deflection patterns, was used to generate speech signals that were presented to listeners in a perceptual test
The purpose was to determine the effect of variations of the magnitude and polarity of the third resonance deflection on identification of the consonant in a V1CV2 disyllable while the deflections of the first and second resonances were held constant
The concept of encoding speech segments at the production planning stage by specifying relative deflections of the first three vocal tract resonances was built into a model developed by Story and Bunton (2017; 2019)
Summary
The concept of encoding speech segments at the production planning stage by specifying relative deflections of the first three vocal tract resonances was built into a model developed by Story and Bunton (2017; 2019). Based in part on the work of Mrayati et al (1988) and Carre et al (2017), the modeling approach relates change in the resonance frequencies to change in vocal tract cross-sectional area through calculations of acoustic sensitivity functions (Fant and Pauli, 1975), which are used to generate speech output. The most recent version of model (Story and Bunton, 2019; ‘SB19’) was demonstrated to produce sentence-level speech, listener sensitivity to variations in specified resonance deflection patterns, or RDPs, has not been investigated. The purpose of this study was to investigate the relation of change in the third resonance deflection to listener identification of the consonants /d/ and /g/ while the deflections of the first and second resonances were held constant.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
