Ebook: A Phonetic Model of English Intonation

A thesis submitted for the degree of Doctor of Philosophy, University of Edinburgh. 1992Abstract
This thesis proposes a phonetic model of English intonation which is a system for linking the
phonological and F descriptions of an utterance.
It is argued that such a model should take the form of a rigorously defined formal system
which does not require any human intuition or expertise to operate. It is also argued that this
model should be capable of both analysis (F to phonology) and synthesis (phonology to F ).
Existing phonetic models are reviewed and it is shown that none meet the specification for the
type of formal model required.
A new phonetic model is presented that has three levels of description: the F level,
the intermediate level and the phonological level. The intermediate level uses the three basic
elements of rise, fall and connection to model F contours. A mathematical equation is specified
for each of these elements so that a continuous F contour can be created from a sequence of
elements. The phonological system uses H and L to describe high and low pitch accents, C to
describe connection elements and B to describe the rises that occur at phrase boundaries. A
fully specified grammar is described which links the intermediate and F levels. A grammar is
specified for linking the phonological and intermediate levels, but this is only partly complete
due to problems with the phonological level of description.
A computer implementation of the model is described. Most of the implementation work
concentrated on the relationship between the intermediate level and the F level. Results are
given showing that the computer analysis system labels F contours quite accurately, but is
significantly worse than a human labeller. It is shown that the synthesis system produces
artificial F contours that are very similar to naturally occurring F contours.
The thesis concludes with some indications of further work and ideas on how the computer
implementation of the model could be of practical benefit in speech synthesis and recognition.