| Text-to-Speech (TTS) system is one to translate given text to speech which can be used in various applications such as information releasing systems, voice response devices, voice services in E-mail and reading machines for the blind. Great progress has been made in the research on Chinese TTS systems and several Chinese TTS systems have been published. However, because of the complexity of Chinese, the current available speech patterns are not very fine. The speech quality of those systems developed from these patterns are not good enough to meet the needs of users.The main purpose of this thesis are to gain a refined prosodic model of Chinese speech. Traditional methods are not used in this thesis and data mining techniques are employed. Data mining is the process of discovering advantageous patterns in database. There are now many data mining algorithms, one of which is neural network. This thesis presents a data mining system using neural network to find useful patterns from Chinese speech database. Study on the tone changes of Chinese two-word phrases has been made and good results have been achieved. They are helpful to develop high quality Chinese TTS systems.The main works carried out in this thesis are as follows:1. Introduce the history and background of speech synthesis and the current research work in Chinese speech synthesis and the prosodic rules.2. Introduce the history of data mining , some important techniques in this area and the process of data mining.3. Introduce the Error Back-Propagation algorithm.4. Introduce the design and implementation of the SpeechDM system. There are five modules in SpeechDM, namely, database management, speech data processing, pattern generation, machine learning and result analyses. These five modules are discussed in this thesis.5. Present the solutions of those important problems of data processing and machine learning, such as how to split Chinese... |
