A study of stress as it relates to computerized speech

ABSTRACT

For computerized speech to be understood, it must be given human qualities, such as stress. This paper discusses how stress is perceived by the listener, primarily through pitch changes. It shows how inflections stress certain words in a sentence, and that a word's internal stress (accent) affects how these inflections are constructed.An implementation of these findings to add stress to computer speech is shown, compared with similar work, and evaluated.The paper concludes that a word's internal stress must be considered in the generation of speech, and that a grammar in the implementation produces satisfying complex inflections.

REFERENCES

• 1: Bollnger, G., "Accent is Predictable (if you're a mind reader)" Language, 48, September 1972, pp. 633-644
• 2: Chomsky, N. and Halle, M., The Sound Pattern of English, 1968, Harper & Row.
• 3: Elovitz, Honey S., et al, "Letter-To- Sound Rules for Automatic Transaction of English Text to Phonetics", IEEE Transactions on Acoustics, Speech, and Signal Processing, December 1976, 24, pp. 446-456.
• 4: Klatt, Dennis H., "Structure of a Phonological Rule Component for a Synthesis-by-rule-Program", IEEE Transactions on Acoustics Speech, and Signal Processing, October 1976, 24, pp. 391-398.
• 5: Lehiste, llse, "Isochrony Reconsidered", J. Phonetics, July 1977, 5, pp. 253-263.
• 6: Lehiste, Ilse, Suprasegmentals, M.I.T. Press, 1970.
• 7: Lieberman, P., Intonation, Perception, and Language, Mass. Institute of Technology Press, 1966.
• 8: Meyer, Derek C., "Construction of Sentence Pause and Intonation in Computer Synthesized Speech", N.C.S.U., 1977.
• 9: Nooteboom, S. G., "The Perceptual Reality of Some Prosodic Durations", J. Phonetics, January 1973, pp. 25-45.