Abstract
Heretofore measurements from sonograms have indicated that the durations of vowels before fortis and lenis consonants exhibit a systematic difference, the latter being longer approximately in a 3 : 2 ratio to the former. Experiments with synthetic speech have shown that vowel duration is at least a sufficient cue for the perception of the fortis-lenis in final position. The role of this cue is evaluated for natural speech signals by the use of digital gating on a PDP-12 digital computer. The stimuli were comprised of English words ending with stops, fricatives, and clusters after tense and lax vowels. Four different durations were gated from the vowel nucleus. The stimuli were gated in four equal intervals at zero crossings to prevent unwanted transients and the stimuli were presented in random order. The method employed was the forced-choice the 50% cross-over points under the different consonant and vowel conditions. In some cases, recognition curves could not be established even after 75% of the vowel was gated. The results of the experiment indicated that vowel duration, voicing, and the duration of silence between the vowel and the final release transient vary in weight as cues under different vowel- and consonant-type conditions.
Published Version (
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