Speech synthesis as a tool for the study of speech production.
| Number | 226 |
|---|---|
| Year | 1977 |
| Drawer | 4 |
| Entry Date | 06/03/1999 |
| Authors | Cooper, F. S., Mermelstein, P., & Nye, P. W. |
| Contact | |
| Publication | In M. Sawashima & F. S. Cooper (Eds.), Dynamic aspects of speech production (pp. 307-322). Tokyo: University of Tokyo Press |
| url | http://www.haskins.yale.edu/Reprints/HL0226.pdf |
| Abstract | Our intent in this paper is to describe some research studies that we are undertaking and to explain our reasons for choosing speech synthesis and the class of research questions that synthesis as a methodology implies. Briefly, we wish to learn what parts of the complex articulatory events of speech production are actually carrying the message, i.e., what articulatory cues the speaker must produce in order that the listener will understand what was said. We think of this as search for the articulatory cues that parallels earlier work we have done on searching for the acoustic cues in speech. There are close parallels between the two kinds of search, and we hae found it useful in planning the work on articulatory cues to draw analogies with our experience in searching for acoustic cues. Hence, we will speak of that experience in presenting our plans. We will even digress into a brief description of a new pattern playback we have built; it will be useful in the planned studies even though it was designed primarily for research on acoustic cues. We have usually spoken of speech synthesis as a tool for the study of speech perception. But the acoustic cues we found all seemed to point back to articulation, implying that we were, in fact, studying production by way of perception. Thus, the parallels between our earlier work and the planned work can be viewed in this way: both were concerned with speech production, though the earlier work was on cues at the acoustic level, whereas the planned work is on cues at the articulatory level. In either case, the distinguishing characteristics of the methodology are that it seeks to find the principal carriers of information, that it tests for these cues by perceptual methods, and that it uses synthetic speech to do so. Obviously, speech is the required stimulus when the perception of a message is to be tested, and synthetic speech has the very great advantage that systematic manipulation of the stimuli is possible, either at the acoustic level or at the articulatory level that precedes it. |
| Notes |
Search Publications