InterSpeech 2021

Downsizing of vocal-tract models to line up variations and reduce manufacturing costs
(3 minutes introduction)

Takayuki Arai (Sophia University, Japan)
Demonstrating vowel production with physical models of the human vocal tract is a part of intuitive education in speech science. The adult male vocal tract was most often used as a model in the past because of the limited availability of physical models, but discussions on different vocal tract sizes were ongoing. Therefore, we focused on downsizing the vocal-tract models in this study, especially the straight models. We reduced the cross-sectional area function for the sliding three-tube model (including the total length) to female adult and child sizes. Furthermore, we created fixed straight models of similar dimensions for the five Japanese vowels. We found that the intelligibility of each model was preserved as long as the ratios of the cross-sectional areas were maintained even if the cross-sections were less than the average human sizes. This indicates that we can reduce the cost of manufacturing the models, as cost is typically a barrier when the models are used for pedagogical purposes.