论文部分内容阅读
维吾尔语音韵律的调节要充分考虑其独特的语言文化,本文研究维吾尔语音的韵律特征并结合重音、停顿等语言现象对维吾尔语的韵律实现调节.本文录制维吾尔语情感语料,应用ANN(Artificial Neural Network)对停顿进行建模计算;重音的调节则依据维吾尔语的词法规则,整理出重音异义词本,将Fujisaki模型应用在局部单词的层面上,更准确的调节重音出现的位置;最后对提取基音频率的方法作改进,提出含调节参数的LPC,实现对语音基音频率曲线的调节.经过测试,模型停顿的计算预判准确率可以达到71.2%,对重音的调节准确度可以达到86.7%,对基音频率的调节可以明显体现出情感调型.
Uyghur language rhythm adjustment should take full account of its unique language and culture, this paper studies the prosodic features of Uyghur language and combines the phenomenon of stress, pause and other language to regulate the rhythm of Uyghur language.This article records the Uyghur language sentiment corpus, the application of ANN (Artificial Neural Network ) To calculate the pause modeling; accent adjustment is based on lexical rules in Uyghur, sorting out the accent words, the Fujisaki model applied to the local word level, more accurate adjustment of accent position; Finally, the extraction The method of pitch frequency is improved and the LPC with adjusting parameters is proposed to adjust the pitch frequency curve of the speech.After testing, the prediction accuracy of model pause can reach 71.2% and the accuracy of accent adjustment can reach 86.7% The adjustment of the pitch frequency can obviously reflect the emotional tone.