Page 30 - Fall2013

Figure 8. Spectrogram of output signals from the bent-type models with sliding blocks (Models A and B)
Figure 7. Experimental set-up for making recordings using bent-type models with sliding bocks (Models A and B)
with sliding blocks were tested: one for front vowels and one for back vowels. As a result, it was confirmed that the five vowels were produced clearly. A single bent-type model with sliding blocks was subsequently designed and it was con- firmed that the five vowels were again produced with reason- able quality. However, it is not easy to slide the blocks in the current model, so improving the design for better usability is a future goal. Ultimately, users should be able to simultane- ously manipulate for tongue height and advancement and to check the actual sounds. One should then objectively evalu- ate the usefulness of this new model in a pedagogical situa- tion in acoustics, speech science, phonetics, etc.
Based on the current author’s teaching experience, one needs both straight-type and bent-type models for different purposes for education in acoustics and speech science. The straight models are much simpler than the bent models, and one can demonstrate that the area as a function of length is the most crucial factor which determines the vowel quality, but not bending itself.
The CT (connected-tube) model, one of the simplest models, is a static model, but if you compare the shapes and the sounds of different types of the CT (connected-tube) model, one can demonstrate that vocal-tract configuration is associated with the quality of vowels. The S3T model, anoth- er simple model that can achieve similar shapes comparing to the different types of the CT model, is a dynamic model, so that a single S3T model can produce different vowels by changing the position and the size of the slider. Thus, this model partially simulates the tongue movement in our vocal tract.
However, the straight models have disadvantages: 1) it is not easy for learners to imagine how the vocal tract is placed in our head; and 2) the tongue movement is less realistic. The bent models, on the other hand, can cover these two points. 1) The vocal tract starts from the throat and ends at the lips, and it is very natural to understand for learners that the vocal tract is bent in between them. Furthermore, 2) more realistic tongue movement can be achieved by the bent-type models. As a result, they can produce a vowel sequence, such as /aia/, with a natural tongue movement. The flexible tongue model with a gel-type material is also a bent-type model, but it is very difficult to reproduce the same vocal-tract configuration multiple times with such model. The bent-type models with
blocks as proposed in this study, however, can easily repro- duce the same configuration repeatedly. In addition, the bent-type models can extend the coverage of sounds into some types of consonant, such as “glides” and “liquids.” In any case, one should select what types of models that one uses for educational purposes depending on what one wants to teach. Therefore, educators will need to have several dif- ferent types of vocal-tract models that they can use for dif- ferent purposes
Acknowledgements
This work was partially supported by Grant-in-Aid for Scientific Research (No. 24501063) from the Japan Society for the Promotion of Science. The author would also like to thank Keiichi Yasu for his assistance.AT
References
1. H. Dudley and T. H. Tarnoczy, “The speaking machine of Wolfgand von Kempelen,” J. Acoust. Soc. Am. 22, 151-166 (1950).
2. W. von Kempelen, Mechanismus der menschlichen Sprache und Beschreibung einer sprechenden Machine, (The mechanism of human speech and the description of a speaking machine), Wien, Austria (1791). (Reprint published in 1970.)
3. T. Arai, “Education system in acoustics of speech production using physical models of the human vocal tract,” Acoust. Sci.Tech. 28, 190-201 (2007).
4. T.Arai,“Educationinacousticsandspeechscienceusingvocal- tract models,” J. Acoust. Soc. Am. 131, 2444-2454 (2012).
5. T. Arai, “Gel-type tongue for a physical model of the human vocal tract as an educational tool in acoustics of speech produc-
Figure 9. Spectrogram of output signals from the bent-type models with sliding blocks (Model C)
Mechanical Models of Vocal Tract 29