Page 66 - Summer2019

Book Review
These reviews of books and other forms of information express the opinions of the individual reviewers and are not
necessarily endorsed by the Editorial Board ofthis Iournal. —Pl1ilip I. Marston, Book Review Editor
A A key issue considered is how a vocal tract model can be
deformed in such a way that results in maximal acoustic varia-
. . ... Authors: Rene Carré, Pierre Divenyi, tion with minimal vocal effort.
I ".-...‘.V.» .. V Moha.mad Mrayati
SPEECH! Publisher: De Gruyter, Berlin, The authors propose a “Distinctive Region Model,” (DRM)
éRl2)Y 'c Germany, 2017, 214 pp. in which “privileged” acoustic trajectories (i.e., formant
I . Price: £79.56 (Hardcover); patterns) serve as “coding primitives.” In their view, infor-
£63.96 (Kindle) mation is encoded in articulatory trajectories rather than
_ ' ISBN: 978-1-501-510601 as a sequence of quasi-static “targets.” “Locus equations,”
which compute the slope and duration of formant tra.n-
The world’s many tongues vary greatly sitions, are shown to reliably distinguish among vowels.
in their sound patterns, lexicons, and Because these trajectories are shaped by the interaction of
grammar. Despite such variation, all languages share cer- vocalic and consonantal context, the framework is essen-
tain properties in common. One of especial interest is the tially one in which the syllable forms a fundamental unit of
intricate interleaving of consonants and vowels in the artic- speech production and perception.
ulatory stream, instantiated as a low-frequency modulation
in the speech waveform. Such fluctuations are the result Although the model is restricted to vowels and stop con-
of articulatory movement of the lips, tongue and jaw, and sonants, it appears to capture much of the speech signals
which serve as the acoustic foundation of a sophisticated dynamic properties. One of the most intriguing properties
information-bearing system. of the DRM is its ability to produce a broad range of speech
sounds with a relatively small number of control para.m-
Why speech is so configured is rarely addressed in the sci- eters, an insight of potential interest to speech clinicians
entific literature. and engineers.
Speech: A Dynamic Process examines the question from A later chapter focuses on perceptual aspects, including
several vantage points through a clever combination of empirical studies of processing dyna.mic speech sounds.
modeling, theory, and empirical studies. This work is then linked back to the modeling and pro-
duction studies discussed in the earlier chapters. A case is
The book begins with a consideration of two approaches to made for the human auditory system’s specialization for
scientific inquiry—induction (with its focus on data col- speech, especially for processing communication signals
lection, statistics, modeling and prediction) and deduction in the presence of background noise and other forms of
(emphasizing “universal” principles, “logic,” formal model- acoustic interference.
ing, and verification). These complementary perspectives are
used to integrate the trifecta of speech production, acoustics, The final chapters are more philosophical in tone, pre-
and perception within a unified theoretical framework. senting the case for a dynamic theory of speech based on
melding the inductive and deductive approaches discussed
This philosophical introduction is followed by an historical earlier. The theory is also examined from the perspectives
review of speech research, with a focus on the latter half of linguistics, psychology, engineering, and evolution-
of the 20th century that sets the scene for the vocal-tract ary theory.
modeling work discussed in later chapters.
Speech: A Dynamic Process is a thought-provoking book
The models range from the primit.ive— a simple tube — to the that examines the speech signal in many interesting ways.
more elaborate, incorporating far greater degrees of freedom. However, the theoretical framework proposed has its
s4 | AA:auuIiI:l‘I'b:lay| Summer 2019