Syllable based text to speech synthesis system for Turkish texts

In this thesis study two different text to speech synthesis systems are designed. In the first system syllable based word synthesis has been practiced. While the second system has been designed as object based and has utilized diphone database as the articulated part. Designed first system generated on MATLAB basis has used Concatenative Synthesis Method. In the method which is based on the idea to articulate the phonemes that form the word by joining them in sequence, syllables are selected as the parts to be concatenated. For this system, it is enabled to record syllables beforehand and to delete unvoiced parts. The following process is passing these segments through Hanning Window and adapting these to other segments frequencies and pitches. At last, using one of the phoneme concatenating algorithms, SOLA (Synchronized Overlap Add Algorithm), sample word synthesis is generated. Second system is generated by using C# programming language. The system that has many specialities consisting in litterateur of articulating Turkish texts can be used successfully in Windows Operating Systems. In the system which is successful in articulating numbers, abbreviations and files, the role of punctuation marks on articulating has also been regarded. Moreover, in this study?s first parts there have been acknowledgements on text to speech systems and systems designed for Turkish language have been examined in detail.