MEASUREMENTS AND MODELING IN SPEECH PRODUCTION RESEARCH
D. H. Whalen · 2004
Early measurements of the vocal tract have led to numerous insights into the way the acoustic signal is shaped by the mouth. One important adjunct to measurement has been analog modeling and detailed acoustic analysis. Some early measurements, such as Stevens’ exemplary x-ray film, are still being analyzed today, due to the continuing challenge of measuring the speech articulators. Although x-ray movies have provided the fullest depiction of the moving vocal tract, they are difficult to obtain and quantify. Recent point source measurements (electromagnetometry and x-ray microbeam) have been successful in characterizing anterior tongue movements, but their inherent limitations prevent imaging of the pharynx, which has a large effect on the acoustic output. Very recently, cine MRI and opticallycorrected ultrasound have surpassed the x-ray capability of obtaining large data sets of images of the vocal tract during running speech. These techniques are less demanding on the speaker and thus applicable to a wider range of talkers. With these new measurements, we should be able to refine our modeling, which always has compromises, and to better explore the relationship between vocal tract shape and the resulting acoustics. They should also help us address the inverse problem (i.e., that there are, theoretically, many possible tract shapes that can result in a given spectrum), and allow us to examine changes in articulator position that do not result in measurable acoustic differences. We are poised to build upon the substantial foundations of speech production research and advance to additional insights with these new techniques.