Contents of the thesis

are organized as follows: Chapter 2 presents briefly the underlying methodology and the background for the thesis. The main algorithms for SOM, LVQ and HMMs are described and their properties central for this work are stressed. The main results and methods obtained by the author are summarized and motivated in Chapter 3. The exact experimental results are presented in the publications which are part of the thesis. The detailed description and formulation of the segmental LVQ3 training method and tuning method based on LVQ2 is given in the publications, as well. Brief conclusions are given in Chapter 4. At the end of this introductory part there is a short summary of the publications and a highlighted list of the main contributions of the author.