基于音高映射合成语音的汉语双字调声调训练

发布时间：2018-03-30 13:13

本文选题：语音教学　切入点：语音习得　出处：《清华大学学报(自然科学版)》2017年02期

【摘要】：该文使用音高映射方法,通过选择合适的标准语音,合成出音段、音色保持不变,而只是声调变为标准语音声调的教学语音用于声调训练,减少了语音信号中声调信息之外的复杂变化带来的信息冗余与干扰。以汉语双字调的合成语音为实验材料,对日本被试进行了声调训练实验。训练结果表明:合成语音方法在声调的感知和产出的相对进步率,以及泛化产出的效果上都优于标准语音方法,远好于没有训练的对照组,大部分实验结果差异在统计上具有显著性。实验结果佐证了语音学习时存在人脑的选择性注意机制,为将合成语音方法集成到计算机辅助汉语声调教学系统,提供了实验和理论依据。
[Abstract]:In this paper, using pitch mapping method, by selecting the appropriate standard speech, the timbre is synthesized, and the tone color remains the same, but only the teaching voice which turns the tone into the standard tone is used for tone training. The information redundancy and interference caused by the complex change of tone information other than tone information in speech signal are reduced. A tone training experiment was conducted on Japanese subjects. The results showed that the synthetic speech method was superior to the standard phonetic method in the perception of tone, the relative improvement rate of output, and the effect of generalization, and was much better than that of the control group without training. Most of the differences in experimental results are statistically significant. The experimental results prove that the selective attention mechanism of human brain exists in phonological learning. In order to integrate the synthetic speech method into the computer-assisted Chinese tone teaching system, The experimental and theoretical basis is provided.
【作者单位】：北京语言大学信息科学学院;北京语言大学对外汉语研究中心;
【基金】：国家自然科学基金项目(61175019) 北京语言大学梧桐创新平台项目(中央高校基本科研业务费专项资金资助)(16PT05)
【分类号】：TN912.33;H116.4-4

【相似文献】