Sixth European Conference on Speech Communication and Technology (EUROSPEECH'99) Budapest, Hungary September 5-9, 1999

A Vocal Tract Model Using Multi-line Equivalent Circuits

Nobuhiro Miki (1), Thoru Yokoyama (1), Takeshi Ohtani (1), Shinobu Masaki (2), Ikuhiro Shimada (3), Ichiro Fujimoto (3), Yuji Nakamura (3)

(1) Graduate School of Engineering, Hokkaido University, Sapporo, Japan
(2) ATR, Kyoto, Japan
(3) Takanohara chuo Hospital, Nara, Japan

In this paper we show a speech production model of the vocal tract analogue, The 3-D vocal tract shape has information for phonetic or personal characteristics, and it has an important meaning to study for the relation between the vocal tract shape and the acoustic characteristics. The FEM analysis is a useful tool for the acoustic analysis, but the computational cost is a problem. On the contrary to the FEM modeling, the transmission line model is convenient to evaluate the impedance function without high computational cost. We propose a new multi-line model which is available to represent the muli-channel as liquid sound. but the involving the 3-D parameters is Difficult because of the one-dimensional model essentially.

Full Paper (PDF)

Bibliographic reference.  Miki, Nobuhiro / Yokoyama, Thoru / Ohtani, Takeshi / Masaki, Shinobu / Shimada, Ikuhiro / Fujimoto, Ichiro / Nakamura, Yuji (1999): "A vocal tract model using multi-line equivalent circuits", In EUROSPEECH'99, 129-132.