利用Praat软件提取塞音、塞擦音的嗓音起始时间VOT特征参数并对此进行了统计分析,藏语拉萨话塞音以及塞擦音文本均选自"藏语单音节声学参数数据库", 实验结果表明利用VOT均值能够明显地把藏语拉萨话塞擦音送气、不送气类别上区分开来.从发音方法上看,塞音和塞擦音的VOT数值有明显差别.同一种发音部位的辅音,由于发音方法不同,VOT的值也明显不同. 研究结果对藏语语音合成自然度和语音识别系统识别率的提高有非常大的作用.%In this paper, by using the software Praat plosive, affricate voice onset time feature parameter extraction and statistical analysis, in Lhasa Tibetan plosive and affricate texts are selected from the group consisting of "Tibetan monosyllabic acoustic parameters database. The experimental results show that using VOT mean can significantly to Lhasa Tibetan affricate aspirated and unaspirated category area separated.From the way of the pronunciation, there are obvious differences between the VOT values of plosives and affricates. The same pronunciation of the consonant, due to the pronunciation method is different, the value of VOT is also significantly different.Results for improving the naturalness of Tibetan speech synthesis and speech recognition system has a very big role.
展开▼