INVPOTFUN · The application of time-domain, inverse-scattering methods to the reconstruction of the vocal tract potential function
FP6 — Marie Curie Actions (Human Resources and Mobility)
- Duration
- 2005-09-01 → 2007-08-31
- EU contribution
- €152,809
- Participants
- 1
- Scheme
- EIF
Lines connect the coordinator with its partners.
Project objective
The main goal of this proposal is the development and application of time-domain, inverse-scattering methods to the reconstruction of unknown potential functions in the one-dimensional acoustical Klein-Gordon equation. This equation is used to model the mechanism of speech. It has been recently shown that the acoustical Klein-Gordon equation precisely determines the natural resonance frequencies (formants) of the human vocal tract.The novelty of this model is that the sound is specified by a unique configuration of barriers and wells along the tract, by analogy with quantum-mechanical wave-scattering problems. This is in contrast to conventional models based on the Webster horn equation and the vocal tract area function, where the same sound can be produced u sing many different vocal tract area functions and little progress has been made over many years.In consequence, this new result opens up the possibility of identification of sounds on the basis of the shape of the vocal-tract potential alone. We intend to develop techniques for the reconstruction of this potential from real human speech. In the preliminary part of the research we will examine different numerical techniques for solving the forward problem - the sound production. We then intend to develop the method of the vocal tract potential inversion from the output pressure-wave data, based on methods developed in reflection seismology.In the last part of the project we intend to apply these techniques to real human vowels. The proposed research has the potential to lead to better speech recognition devices then existing ones which currently do not take vocal tract physics into account and cannot adapt to natural human speech styles, such as ordinary conversation.
Original text from CORDIS.
Participants
- KING'S COLLEGE LONDON · LONDONCoordinatorUnited Kingdom
Links
Data: CORDIS, © European Union
