TubeTalker is a moniker for the area function model described in: Story, B.H., (2005). A parametric model of the vocal tract area function for vowel and consonant simulation, J. Acoust. Soc. Am., 117(5), 3231-3254. The model is more recently described in Story, B.H., (2011), Proceedings of the International Workshop on Performative Speech and Singing Synthesis March 14-15, 2011, Vancouver, BC.
The purpose of developing this type of speech synthesizer is to provide a tool that can be used to facilitate understanding human production of speech and singing. The intent is not necessarily to produce a synthetic speaker for purposes of text-to-speech systems or other speech technology applications. Nonetheless, our goal is to advance this model such that it produces high-quality, intelligible speech.
The audio demonstrations on this website were produced with TubeTalker coupled to a digital waveguide model of the vocal tract. The voice source is a kinematic model of the 3D glottis based on Titze (2006), The myoelasticaerodynamic theory of phonation, NCVS, pp. 197-214. The basic system of wave propagation is described in my dissertation.
A more complete explanation and additional demos of TubeTalker are available in Story (2013), Phrase-level speech simulation with an airway modulation model of speech production, Computer Speech and Language.The article can be accessed here . Click on this link for audio samples that coincide with this paper.
The movies below are similar to those included with the article, except that these contain a synchronized audio track. In each case, the vocal tract is shown in a pseudo-midsagittal configuration. That is, the shape of the airway extending from just above the vocal folds to lips is shown along a curve based on anatomical considerations, but no anatomical structures are indicated.