SpeechDM: Learning Patterns from Two-word Phrases in Chinese Speech Database

Pitch model is very important for speech synthesis, and it mainly describes the variation of pitch. The models that are now being used in Chinese speech synthesis are described qualitatively and with low precision. We try to find the pitch model from actual speech samples by data mining. A prototype called SpeechDM has been implemented to learn the pattern of the pitch variation in Chinese two-word phrase. This paper first gives the architecture of SpeechDM, then the learning modules are introduced in detail, and the results are also given at last.