Potential for Estimating Age-Dependent Degradation using the Acoustic Characteristics of Daily Conversation
Potential for Estimating Age-Dependent Degradation using the Acoustic Characteristics of Daily Conversation
复制标题
使用日常对话的声学特征估计年龄依赖性退化的潜力
DOI：
10.1109/gcce.2018.8574840
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Nakatoh Yoshihisa
中科院分区：
文献类型：
--
作者：
Wakita Yumi;Kondo Shinya;Nakatoh Yoshihisa
我们的目标是建立一个系统，根据老年人的日常对话来估计他们年龄依赖的退化程度。我们已经证实，基频(F0)的时间相关特征对于估计对话流畅性是有效的。在这里，我们调查了这些特征是否也可以有效地估计年龄相关的退化。通过对自由会话语音数据的分析，我们证实了当语音印象小于75时，每个话语的F0的标准偏差值与F0的平均值相关。这种相关性在年龄较大的个体中较弱。我们的结果表明，上述特征的F0将是有效的估计年龄相关的退化。
Our aim is to create a system to estimate the degree of age-dependent degradation in elderly persons based on their daily conversations. We have already confirmed that the time-dependent features of the fundamental frequency (F0) are effective for estimating conversation smoothness. Here we investigated whether these features are also effective for estimating age-dependent degradation. By analyzing free conversational speech data, we confirmed that the standard deviation value of F0 for each utterance is correlated to the average value of F0 when the impression of the speech is regarded as under 75. This correlation was weaker in older individuals. Our results indicate that the above features of F0 would be effective for estimating age-dependent degradation.