The Text-to-Speech Problem

The process of reading aloud then is one of taking a signal of one type, writing, and converting it into a signal in another type, speech. The questions we will explore in this chapter are  

It is clear that natural (human) language developed first using sound as its medium. While humans clearly also communicate with facial and hand gestures and so on, these systems are not able to express the same wide range of expressions as natural language. Quite why human language 

To begin with, we need to look at the nature of speech and writing in more depth. We will do this from a number of angles, looking first at the physical differences between these signals, before moving on to look at how these differences infiuence communication. 

While a few problems certainly occur, this model seems to fit the facts very well. Whether we write or say the sentence the man walked down the street, there are six words and they occur in the same order in the spoken and written versions. This fact may be so obvious as to seem superfluous, but it is in fact vital to our discussion. One would find it difficult for instance to teach someone to read French text and speak English words even if we learned that the pronunciation for chien was /d ao g/, we would find it difficult to understand why the written French sentence has different numbers of words in different orders from the spoken English version. 

This conclusion is so important for our system that it is worth restating again by and large, the identity and order of the words to be spoken is all we require to synthesise speech no higher order analysis or understanding is necessary. 

---

Before delving into the details of how to perform text-to-speech conversion, we will first examine some of the fundamentals of communication in general. This chapter looks at the various ways in which people communicate and how communication varies depending on the situation and the means which are used. From this we can develop a general model which will then help us specify the text-to-speech problem more exactly in the following chapter. 

---