Natural Speech Synthesıs For Turkısh Text-To- Speech Conversıon
In this thesis, a unique text-to-speech synthesizer for Turkish is presented. The system follows the concatenative method which is contemporarily used in text-to-speech conversion systems. In this approach, the speech segment unit is defined as “at most two letter syllable” and the system is built accordingly. The thesis can be considered in two parts as follows:
In the first part of the thesis, the relationship between the physiology of the human speech system and the acoustics of the speech is investigated. Then, the advancements and different methodologies on speech synthesis for more than two centuries are summarized. Finally, contemporary text-to-speech synthesizers are discussed with comparisons.
In the second part of the thesis, the framework and methodology of the proposed Turkish text-to-speech synthesizer are explained. The methodology of the system is also compared to contemporary prevailing text-to-speech synthesis methodologies. Afterwards, the positive and negative aspects of the presented system is evaluated, and the system’s contribution to text-to-speech synthesis for Turkish is presented. Lastly, some approaches are proposed for both the text-to-speech synthesizer presented in the thesis and text-to-speech synthesizers in general. The program codes of the presented system, which is built on MATLAB software, is also attached as an appendix to the thesis.
NOORİ Abbas
Danışman : Yrd. Doç. Dr. Fırat KAÇAR
Anabilim Dalı : Elektrik-Elektronik Mühendisliği
Mezuniyet Yılı : 2011
Tez Savunma Jürisi : Yrd. Doç. Dr. Fırat KAÇAR
Prof. Dr. Ayten KUNTMAN
Prof. Dr. Sıddık YARMAN
Prof. Dr. Aydın AKAN
Prof. Dr. Ahmet SERTBAŞ
Dostları ilə paylaş: |
