haskins logo
(This section is based, in part, on a paper by Philip Rubin and Eric Vatikiotis-Bateson, entitled,
"Measuring and modeling speech production", that appears as a chapter in: S.L. Hopp, M.J. Owren, and C.S. Evans (Eds)., Animal Acoustic Communication. Springer-Verlag, 1998.)


In human communication, the speech system is specialized for the rapid transfer of information (Liberman et al., 1967; Mattingly & Liberman, 1988). Significant events in the acoustic signal can occur in an overlapped or parallel fashion due to the coproduction of speech gestures. A result of this is that aspects of the signal corresponding to different linguistic units, such as consonants and vowels, often cannot be isolated in the acoustic stream. One way to help tease apart the components of the speech signal is to consider the physical system that gives rise to the acoustic information: the acoustic encoding of phonetic information is then viewed in light of the flexibility inherent in the production apparatus, particularly the human supralaryngeal vocal tract, in which individual articulators or groups of articulators can function semi-independently. In this section we will review this approach. First, we show how the analysis of speech acoustics has benefited by treating the sound production system as one in which the contributions of physical acoustic sources and physiologically-determined filters are combined. We then discuss how acoustic diversity has resulted in a desire to find articulatory simplicity. In the process, we review some of the methods used to examine articulatory activity, and also describe in detail a particular attempt at modeling the coordination of the speech articulators. We close by considering some recent attempts to explore the links between production, perception, and acoustics . Where possible, recent trends in the field have been exemplified by projects involving ourselves and our colleagues.

In order to consider the details of acoustics and production, it is necessary first to briefly describe the system that we are studying. Figure 1 shows a schematized view of the human sound production system. Note the extent of the anatomy which is involved in sound production (Lieberman & Blumstein, 1988; Borden & Harris, 1984; Ladefoged, 1975). There are variety of ways to produce sound. One method involves using the air pressure provided by the lungs to set the elastic vocal folds into vibratory motion. The larynx converts the steady flow of air produced by the subglottal system into a series of puffs, resulting in a quasi-periodic sound wave. Aperiodic sounds are produced by allowing air to pass through the open glottis into the upper airway (the supralaryngeal vocal tract) where localized turbulence can be produced at constrictions in the tract. A third method involves producing transient clicks and pops by rapid release of the articulatory closure (Ladefoged, 1975). Here the sound sources arise from the local changes in the vocal tract and do not require air pressure from the subglottal system.

Figure 1

Figure 1: The human speech production system.

The Acoustic Theory of Speech Production
Measuring and Analyzing Speech Production
A Model of the Human Vocal Tract
Gestural Modeling
The State of the Art