摘要： 介绍了语料库设计的原理和过程,并结合粤语语音合成系统语料库实现,提出了“语境矢量”的独特设计和“语境总量”的概念,给出了语境总量的计算方法。最后介绍了Greedy算法,并给出了文本语料搜索的具体流程。

关键词: 语音合成, 语料库, 语境矢量, 语境总量

Abstract: This paper describes the corpus design of a Cantonese text-to-speech system, and gives an introduction of the phonetic knowledge ofCantonese, then proposes the unique design of Cantonese “context vector”(CV) and the unique concept “context gross”(CG), along with thecalculation method of CG, laying a theoretical foundation for text research. After that, by carrying the Greedy algorithm, it gives the process of thecorpus design.

中图分类号: 

• TP391.42