Xxviii. Speech Communication Academic and Research Staff A. Real-time Spectral Input System for Computer Analysis of Speech

The objective of the research in speech communication is to gain an understanding of the processes whereby (a) discrete linguistic entities are encoded into speech by human talkers, and (b) speech signals are decoded into meaningful linguistic units by human listeners. Our general approach is to formulate theories or hypotheses regarding certain aspects of the speech processes, obtain experimental data to verify these hypotheses, and simulate models of the processes and compare the performances of the models and of human talkers or listeners. Research in progress or recently completed includes: observations of the acoustic and articulatory aspects of speech production in English and in other languages through spectrographic analysis; study of cineradiographic data and measurement of air-flow events; study of the perception of speech sounds by children and examination of the acoustic properties of the utterances of children; computer simulation of articulatory movements in speech; investigation of the mechanism of larynx operation through computer modeling and acoustic analysis; examination of new procedures for analysis of speech signals using deconvolution techniques; experimental studies of the perception of vowel sounds; speech synthesis by rule with a computer-simulated terminal analog synthesizer; a re-examination of the system of features used to describe the phonetic segments of language; and the development and improvement of interface equipment for spectral analysis of speech with a computer and for synthesis of speech from computer-generated control signals. K. N. Stevens, M. Halle