Mechanical modeling of speech production: Learning from the past and going beyond

The speaking machine designed by von Kempelen in the 18th century was one of the most famous mechanical models of speech production. Although the systematic configuration of the vocal tract was not replicated by this machine, there were different sound sources for vowels and consonants. In the early 20th century, Chiba and Kajiyama measured and reproduced the human vocal tract, of which physical models produced sounds resembling naturally produced vowels. Later in the 20th century, Umeda and Teranishi investigated vowel and voice qualities with a mechanical model of the human vocal tract by inserting bars into an acoustic cavity. My educational tools for speech production are applied and/or inspired by all of these models and now used in my classroom teaching. Some of the models are further used for research purposes. One of them is for investigating the perceptual interaction between voice quality and nasality. The advantage of this mechanical model is intuitiveness of understanding, especially the relation between manipulating parameters and speech outcomes. Thus, mechanical models are still useful along with computer models. Work supported by JSPS KAKENHI Grant No. 24K06423.