A SIMPLE LINEAR MODEL FOR VOWEL PERCEPTION,
CALIFORNIA UNIV LOS ANGELES
Pagination or Media Count:
This work had two major aims. One was to obtain sufficient data to make reliable estimates of confusions among vowels after ideal filtering. The other was to compare the responses of the same listeners to real vowels and their synthetic aliases. Beyond their inherent interest, these data are valuable in assessing vocal analog systems whether as practical devices or as research tools and speech standards. A simple, quantitative model is presented and tested against the perceptual data from real and synthetic talkers.