The Tilt model

The Tilt intonation model developed by Taylor [434] [439] [436] was developed with the explicit intent of creating a practical, engineering model of intonation. To this extent, issues of pmely linguistic concern (such as phonological rules in the AM model) or biological plausibility, as in the 

Fujisaki model were ignored. While the Tilt model is no more plausible as a means of intonation production that say concatenative synthesis, it was designed so that its parameters would have a clear linguistic interpretation. 

The Tilt parameter itself is used to define the general shape of the event, independent of its amplitude, duration or alignment. To find the value of Tilt, we first calculate an interim value. Tilt amplitude, found by taking the ratio of the differences of the rise and fall amplitudes divide by 

Rgure 9.12 A pitch accent split into rise and fall components. 

Amphtude = A = l rfeel + Mraul Duration = D = [Drisel + Dfau  

These two quantities are highly correlated, which allows us to combine them into the single final Tilt parameter  

Rgure 9.13 Five pitch accents with differing values of tilt, from +1 (pure rise) to — 1 (pure fall). 

---

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. 

Articles on all the models and techniques described here are readily available, but there are few which consider all models together and offer comparisons. The best account of the AM model is given by Ladd [269]. Ladd not only describes the model and its developmenL but also serves as a solid introduction to the whole field of intonational phonology and the practical phonetics associated with it. Accounts of the INTSINT model are given in Hirst [212], the Tilt model in Taylor [436], the Fujisaki model in Fujisaki [166], the SFC model in Bailly and Holm [31], the Dutch model in [447], the British school in O Connor and Arnold [333]. 

The Tilt model describes intonation with a sequence of events, each of which is described by a set of continuous parameters. 

Taylor, P. A. Analysis and synthesis of intonation using the tilt model. Journal of the Acoustical Society of America 107, 4 (2000), 1697-1714. 

In the Tilt model, there are two types of event, accent and boundary (fliese are effectively the same as the pitch accents and boundary tones in the AM model). Each event comprises two parts, a rise and a fall, as shown in Figure 9.12. Between events straight lines called connections are used. All variation in accent and boundary shape is governed firstly by the relative sizes of the rise and fall components and secondly by how the event aligns with the verbal component (basically whether it is higher or... 

---