Consonant-Vowel Ratio and Speaker Intelligibility
Carl E. Williams, B. Woods, Michael H. L. Hecker, Kenneth Noble Stevens · The Journal of the Acoustical Society of America · 1966
Since voiced and voiceless speech sounds are generated by different mechanisms, it might be hypothesized that differences in speaker intelligibility are due in part to differences in the production of these sounds, and, in particular, are related to the consonant-vowel (C/V) ratio for voiceless consonants. To explore this question, eight selected male speakers recorded lists of the Modified Rhyme Test; two of the speakers made recordings at three levels of vocal effort. Initial and final /s/-vowel and /t/-vowel ratios were obtained from graphic-level tracings of the recorded utterances. The lists were presented to listeners at two signal-to-noise ratios. Significant correlations were found between speaker intelligibility and final C/V ratios for the consonants examined. There were monotonic relations between vocal effort and /s/-vowel ratio and between /s/-vowel ratio and intelligibility. [Work supported by Decision Sciences Laboratory, Electronic Systems Division, U. S. Air Force Systems Command, Bedford, Massachusetts.]