TABLE OF CONTENTS Generally the aim of researchers is to extend knowledge and thereby facilitate progress. The key to achieving this is flexibility and a supply of good ideas. But often the prerequisite cannot be realized because too much emphasis is placed on what has been achieved and what is presently available. Often there is not enough flexibility and courage when it comes to going right back to earlier steps or even to the beginning to deal with the basic issues once again, but from a different perspective.
This is also the case with speech technologies: If we systematically examine the pros and cons of all the possible system structures, the bases of our knowledge and process control, it will become clear how we should construct the best possible system to deal with today's requirements. However, already at 1984 Fant regrets that activities concerned with speech synthesis are all too often concerned with implementing a functional system and that enhancing knowledge is less important [67]. This is often the case still today. Pointedly going back to the basics and dealing once again with specific issues can increase the necessity to find new solutions. This can have far-reaching consequences for system performance, e.g. for the quality of synthetic speech: Basic changes can, at first, lead to a considerable loss of quality of the system output, which, however, in the long term will facilitate progress. What is a loss of quality for the system output, may...
