Skip to main content

Physiological models of speech production

Objective



Research objectives and content The object of the proposed research is to develop and experimentally validate a physiological model of speech production, i.e. a detailed and realistic biomechanical model of the whole oral cavity (including the larynx, the tongue, the jaw, the hyold bone,and the lips). The proposal is motivated by evidence that the complex mechanics of the oral cavity has to be accounted for when investigating the mechanisms underlying the neural control of speech movements. To this purpose, the model needs to be highly realistic from the mechanical point of view, and, in particular, the physiological and biomechanical contributions to the movement must be modelled in detail: namely, tissue mechanics, the dynamics of muscle contraction and the role of peripheral circuitry. The model will be experimentally validated and then used to explore a number of issues related to the neural control of speech production. More specifically, an hypothesis of 'simplicity' of the central control commands will be tested by inferring their structure for simple speech sequences (by comparison with empirical data, and application of dynamic optimisation techniques), and by investigating how they need to be modified to deal with different speak rates. Likewise, I will address the control mechanisms underlying the coordination of different parts of the oral cavity, as well as the phenomenon of coarticulation. It is clear, however, that the range of possible applications in speech research of a 'virtual speaker' or 'talking robot'is countless. Training content (objective, benefit and expected impact) The proposed research will allow a systematic analysis of speech events in terms of biomechanics. As for myself, it will help to improve my competence in the experimental techniques of motion analysis that are specific of speech. Moreover, the proposed methodological approach (namely, the simulation techniques and the experimental paradigms) is also directly applicable to other domains of human movement science, and also to neuromorphic robot control. Links with industry / industrial relevance (22) Although the project has no immediate industrial relevance, a model of such kind may potentially have a wide range of research, educational and technological applications, including work in articulatory speech synthesis, automatic speech recognition, computer animation, and virtual surgery,

Funding Scheme

RGI - Research grants (individual fellowships)

Coordinator

Institut National Polytechnique de Grenoble
Address
46 Avenue Félix Viallet
38031 Grenoble
France