Experimental Study on Artificial Vocalization by Using Artificial Vocal Chord-Vocal Tract System. and Theoretical Approaches on Vocalization Mechanica of Vocal Chord-Vocal Tract System

建立了用于分析超弹性有限变形土的本征频率和动力响应的三维有限元系统。结合从肺到声道的潜流有限元系统。研制了一种解决声带自激振动的整体系统。并应用于求解声带自激振动中的各种频率，这些频率随声带内部肌肉的激活而变化。通过与22岁活人声音的和弦振动测量结果和FFT测量结果的比较，表明上述方法得到的仿真结果可以很好地应用于解决该领域的问题。对舌缩至硬腭分离的多连通口腔的声学共振方程进行解析求解，得到各五个元音的共振峰频率。在此基础上，提出了用正交样条函数表示声道三维结构的一般方法。结合上述方法，利用基于瞬态声场有限元法的多步最小二乘方法，确定了各元音的声道共振形状。将上述方法得到的各元音频谱与22岁的实际人的元音频谱进行了比较，得到了令人满意的一致性。其次，通过使用人工发音装置，以同样的方式进行了一系列人工辅音发声测试。从这些结果中发现，每个辅音谱。最初产生于收缩区。尽管从辅音头部到后元音的转换过程不同，尽管后元音的类型不同，但它们的频谱几乎相同。最后，利用K-epsilon模型进行了口腔收缩区附近的湍流流动分析，以考察其发声特征，并将所得结果与上述最小二乘法的结果进行了比较，得到了较好的相关性。少

Three-dimensional finite element system for analysing eigen-frequency and dynamic response of hyperelastic finite-deformation doby have been developed.Combined with the finite element system of the potential-flow for beathing flow from lung to vocal-tract. a total system to solve the self-exciling vibration of vocal chord has been developed. and applied to solve a wide varieties of frequencies in the self-exciting vibration of vocal-chord, which vary with the activation of interior muscles of the vocal-chord.It is demonstrated, through comparisons with mesurement results of the vocal-chord vibrations and the FFT in voice of living human of 22 years olds, that the simulation result, obtained by the above method, can be satisfactorily applied to solve this fields of problems.The acoustically resonant equations of plurally connected mouth-cavities, which is separated by tongue-contraction to hard palate are analytically solved for obtaining Formant-frequencies of respective five-vowels. T … More hen the general method to express the three dimensional configuration of the vocal-tract by using orthogronal spline functions has been developed. Combined with above methods, resonant shapes of the vocal-tract for respective vowels have been identified by using multi-steps of least spuare-approaches, based on the finite element method for transient acoustical-fields. The frequency spectors for respective vowels, obtained by above approaches, are compared with those of actual human of 22 years olds, and satisfactorily good concordances between them are obtained.In the next place, a series of artificial consonant vocalizing tests has been carried out in the same way by using artificial pronounciation devices.It is found from these results that each consonant spector. initially generated in the contraction zone. has nearly common frequeucy spector, despite of different transition process from consonant head-part to suceeding vowel, and despite of the type of succeding vowel.Finally, the turblent flow analysis near contracting zone in mouth cavity, has been carried out by using K-epsilon model, in order to review the characteristics of the vocalization, and the obtained results are compared with those of above least square approach, yielding to comparatively good correlations. Less