Intonational form

The models differ significantly in what they take as the primary form of intonation. In the AM model this is quite abstract while in the Tilt model this is quite literal or acoustie . These differences in primary form should not be taken to mean that the proponents of these models do not believe that there should be more abstract or more conerete representations, just that the best representation happens to lie where they describe it. In the many synthesis schemes based on the AM model there are other, more phonetic or acoustic levels, and in the Tilt model there is always the intention that it should serve as the phonetie description of some more abstract higher level representation. 

---

The majority of this chapter focuses on the synthesis of intonation. The main acoustic representation of intonation is fundamental frequency (FO), such that intonation is often defined as the manipulation of FO for commimicative or linguistic purposes. As we shall see, techniques for S5mthesizing FO contours are inherently linked to the model of intonation used, and so the whole topic of intonation, including theories, models and FO synthesis is dealt with here. In addition, we cover the topic of predicting intonation form from text, which was deferred from Chapter 6 as we first require an understanding of intonational phenomena theories and models before explaining this. 

In practice standard regression algorithms are difficult to use directly for FO prediction, because of a serious mismatch in the nature of the input and output representations. In general the input representation is a representation of intonational form as just discusses (for example, pitch accent types and positions) whereas the output is a continuous list of real valued numbers. In particular, the feature combination needs to generate not one, but a sequence of FO values, which is further complicated because the number of values in this sequence can vary from case to case ... 

Furthermore, when we come to performing a bottom up analysis, where we wish to describe the continuous contour with a discrete set of linguistic units, we find the mapping even more complicated. Because of these problems, it is nearly universal to use some sort of acoustic model which acts as an interim representation between the abstract intonational form and the FO contour. This model has a fixed number of parameters per unit, and so we can use one of the standard mapping algorithms (such as CART) to generate these parameters from the intonational form ... 

We now turn our attention to this issue of intonation synthesis itself. In Chapter 6 we described a variety of techniques for generating the prosodic form representations of phrasing and prominence from text. We only made passing reference to the issue of generating intonational form fi-om text as this really required a thorough discussion on the issue of intonational form first. With this job now done, we can describe techniques for both generating intonational form descriptions fi om text and for generating FO controls from intonational form. 

Given a representation of intonational form predicted from text, we next attempt to synthesize an FO contour from this. In this section we will examine two rule based deterministic proposals for doing this, the first developed for use the with original Pierrehumbert model and the second, developed more specifically for use with ToBI. 

Applicability of the first approach suggested by Keiko and Zarod-nyuk is based on the unity of thermodynamics and kinetics which explain differently the same physical regularities. As was said above this unity was brilliantly revealed by Boltzmann in his "kinetic" and "thermodynamic" explanations of the second law. In our case, setting, for example, a constraint on the equilibrium constant value of an individual reaction S VjXj = 0 within complex chemical process and writing this constraint intone of the possible forms ... 

Informally we can describe prosody as the part of human communication which expresses emotion, emphasises words, shows speaker attitude, breaks a sentence into phrases, governs sentence rhythm and controls the intonation, pitch or tune of the utterance. This chapter describes how to predict prosodic form from the text while Chapter 9 goes on to describe how to synthesize the acoustics of prosodic from these form representations. In this chapter we first we introduce the various manifestations of prosody in terms of phrasing, prominence and intonation. Next we go on to describe how prosody is used in communication, and in particular explain why this has a much more direct affect on the final speech patterns than with verbal communication. Finally we describe techniques for predicting what prosody should be generated from a text input. 

It is usual to consider three basic components of prosodic form phrasing, prominence and intonation. 

In the model, intonation is composed of two types of components, the phrase and the accent. The input to the model is in the form of impulses, used to produce phrase shapes, and step functions which produce accent shapes. 

The history of intonation research is full of attempts at finding a system of prosodic form which could act as the equivalent of phonemes in the verbal component of language... 

The AM school (which includes ToBI) describes intonation in terms of abstract High and Low tones. Diacritics (, %, -) are used to specify which tones align with syllables and boundaries. The tones can be combined in various ways to form an inventory of pitch accents (e.g. H +L). 

Some models, such as SFC, do away even with notions of explicit prosodic form and learn the all the processes and representations of intonation from data. 

To demonstrate the issue of choice of feature system, let us consider how to represent intonation in the specification. In second generation systems and in some unit selection systems, this is represented solely by the FO contour (sampled in some appropriate way). But as we saw in Sections 9.4, 9.5, when FO contours are generated, they are generated from other features that have been previously calculated. As we cannot actually generate new information, the FO contour is clearly just a transformation of the information contained in the original features into a different form. Hence there is no a priori reason why we can t just use the features we use to calculate the FO contour directly and forgo any actual FO calculation. 

---