|
|
Preface
|
|
|
|
|
|
bling subsets of the overall system, and the provision of a variety of displays to
|
|
|
view the functions of various modules. This system has been used successfully on
|
|
|
several occasions, and is described in Chapter 14.
|
|
|
|
|
|
With this new UNIX version of MITalk, J. Allen and R. Armstrong have un-
|
|
|
dertaken extensive writing and editing which build on the 1979 summer course
|
|
|
notes in order to construct the current text. In particular, all examples are a direct
|
|
|
result of the current implementation, and new rule formulations have been added to
|
|
|
the text by using a generalized notation for phonological rules. These rule im-
|
|
|
provements have been achieved by R. Armstrong. Several new sections have been
|
|
|
added, and extensive editing has been performed along with an expanded and more
|
|
|
explicit representation of the actual algorithms and rules used in the system. Thus,
|
|
|
the present text is the product of the original authors of the 1979 summer course
|
|
|
notes (mentioned above), plus expansion in detail, examples, and both explicit and
|
|
|
extensive rule formulations added by J. Allen and R. Armstrong. M. S. Hunnicutt,
|
|
|
D. H. Klatt, and D. Pisoni have reviewed these changes for accuracy, and the ex-
|
|
|
tensive formatting necessary to produce the camera-ready copy for this book was
|
|
|
done by R. Armstrong.
|
|
|
|
|
|
It is a pleasure to acknowledge the several sponsors of this work over the
|
|
|
years. In the early stages, research was sponsored by the Joint Services Electronics
|
|
|
Program, as well as the National Institutes of Health. For many years, continuing
|
|
|
and generous support has been provided by an anonymous individual donor, sup-
|
|
|
plying the flexibility necessary to pursue appropriate research directions. The four
|
|
|
years of concentrated effort which led to the system’s 1979 version was supported
|
|
|
by the National Science Foundation. It is important to note the donation of a
|
|
|
hardware pitch detector from MIT’s Lincoln Laboratory, designed and built by
|
|
|
T. Bially. The detector was instrumental in providing the very large volume of
|
|
|
pitch contours used as the database to construct fundamental frequency rules.
|
|
|
|
|
|
The MITalk system is the result of an exciting and satisfying project. Much
|
|
|
important research has been performed as a result of its needs, and the overall sys-
|
|
|
tem is an impressive statement of our knowledge in this field. Certainly, there is
|
|
|
still more that needs to be done in order to provide highly natural speech in dis-
|
|
|
course environments. But, MITalk’s contributions are likely to play an essential
|
|
|
role in any of these continuing developments.
|