ASA 127th Meeting M.I.T. 1994 June 6-10

4aED5. Degraded synthetic speech for classroom demonstration.

William J. Strong

Dept. of Phys. and Astron., Brigham Young Univ., Provo, UT 84602

Purposely degraded speech can be useful for demonstrating parameters important in speech perception as well as the effects of common degradations. A four-pole terminal-analog synthesizer whose control parameters could be modified was used to generate natural-sounding and degraded synthetic speech. One category of degradation involved the selective elimination of formants to demonstrate their perceptual role. Another category involved modifying the fundamental frequency contour to change the sense of a sentence or to change its ``expressiveness.'' Another category involved changing the formant frequencies to produce ``helium speech'' or ``argon speech.'' A final category involved degrading the speech waveform via additive noise, bandwidth limitation, peak clipping, and center clipping to demonstrate their effects on speech intelligibility and naturalness. A short demonstration tape will be presented. (This material represents a significant refinement and updating of material originally presented at the 78th Meeting of the Society in November 1969.)