Development of an Intelligent Artificial Larynx

在第一年（1986年）对声门声源的语言学、生理学和声学特性进行初步研究的基础上，第二年（1987年）设计了生物信号检测、控制信号编译和声门声源产生三部分的人工喉，并制作了样机。第三年（1988年），在一个非手术的实验装置上对这些部件进行了组装，以评价它们的组合特性。本项目提出的人工喉作为喉切除病人的替代声源，旨在提供自动控制语音浊声源主要参数的能力，如声源的启停、清辅音时声带振动的瞬时停止、语音基频和振幅的韵律信息变化等参数，其中声源的启停和声带振动的瞬时停止，探讨了用专门设计的开关装置进行手动控制的可能性，同时，还调查了其时间安排的基本性质。这种定时开关在实践阶段将被参与发声的肌肉的剩余功能所取代。通过第三年的实验结果，指出了人工喉在语音质量、清晰度、使用方便性、操作准确性等方面存在的缺陷，并讨论了需要改进的地方。

Based on the preliminary investigation in the first year (1986) on linguistic, physiological and acoustical characteristics of the glottal sound source, three parts of the artificial larynx, namely, detection of biological signal, compilation of control signal and generation of glottal source, were designed and their prototype models were built in the second year (1987). Then, in the third year (1988), those parts were assembled to evaluate the combined characteristics in an experimental set up without surgical operation for the subjects.In the artificial larynx which is proposed in this project as an substitutive voice source for laryngectomized patients, it is aimed to provide capabilities of controlling automatically major paramenters of voiced sound source of speech, such as start and stop of voice source, instant stop of vocal fold vibration at unvoiced consonant, and changes in fundamental frequency and amplitude of voice for prosodic information.As for the start and stop of voice source and the instant stop of vocal fold vibration among those paramenters, possibility of manual control with a specially designed switch device was examined, at the same time, the basic nature of their timing was investigated. This timing switch is to be replaced in the practical stage by residual function of muscules involved in voicing. Possibility of controlling prosodic information conveyed by changes in fundamental frequency and amplitude by using expiratory airflow was also ascertained.As results of those experiments conducted in the third year, defects in the artificial alrynx regarding to quality and clearness of speech sound, easiness of use, and accuracy of operation were remarked, and points to be improved were discussed.