Development of para-linguistic information processing systems based on Gestalt perception and understanding of behaviors of individuals with developmental disorder

引入语音结构的动态特征，用于提高单词识别性能。2)语音结构被用于LVCSR系统中的重新排序过程中，它被发现可以提高性能。3)利用语音结构对幼儿语言习得过程进行建模研究。4)本文提出了一种解释口语产生的理论，即从声音处理的角度考虑人与猿的认知差异。5)本研究采用纵向研究的方法，旨在指导自闭症儿童如何与他人进行充分的社会交往。我们的治疗方法被发现在提高他们的社交技能方面是有效的。

Dynamic features of speech structure are introduced and used for improving word recognition performance. 2)Speech structure is used in a re-ranking process in LVCSR systems and it is found to improve the performance. 3) Modeling of infants’ language ac- quisition process is investigated by using speech structure. 4) A theory is proposed to ex- plain the birth of spoken language, where cognitive differences between human and ape are well considered in terms of sound processing. 5) A longitudinal study is carried out to teach autistic children how to interact with others socially adequately. Our method of therapy is found to be effective in improving their social skills.