保留过渡段信息的MPLPC汉语语音合成方法研究  

A Study of the Speech Synthesis by MPLPC in Standard Chinese Based on Keeping the Information in the Transient Sections

作　　者：谭志强[1] 徐近霈[1] 

机构地区：[1]哈尔滨工业大学模式识别与信息处理教研室

出　　处：《哈尔滨工业大学学报》1994年第4期66-70,共5页Journal of Harbin Institute of Technology

基　　金：青年基金

摘　　要：在已研制的ＭＰＬＰＣ规则合成系统的分析与改进的基础上，提出了采用保留语音相邻音素间过渡段信息为合成单元，对稳态音素进行帧压缩，用ＭＰＬＰＣ和音调调制，按规则合成较高质量的汉语单音节语音方法。This paper presents a high quality real time Chinese monosyllable speech synthesis by rule method. The method uses transient sections between two adjacent phonemes to abtain high speech, compresses stable phoneme to one segment for data reduction, and employs Multi-Pules Linear Prediction Coefficients (MPLPC) as parameters. Pitch period, tone, energy and the length of voice can be controlled flexibly to generate required speech. Hardware and software structures and basic data base generation are also introduced.

关 键 词：语音合成 汉语 MPLPC规则 

分 类 号：TN912.33[电子电信—通信与信息系统]