Application
of Neural Networks and Pattern Recognition Methods to the Evaluation
of Speech Deformation Degree for Patients Surgically Treated
for Larynx Cancer by Elzbieta Reron, Ryszard Tadeusiewicz, Maciej Modrzejewski
& Wieslaw Wszolek Key words: cancer of the larynx; voice quality; neural networks
Abstract:
The
work presented concerns the automatic evaluation of pathological
speech, and finding the methods of objective acoustic evaluation
of the degree of speech deformation. The most interesting techniques
are neural networks, which adaptation and learning ability is
well known, and realization of which is now possible in terms
of equipment. The objective of this presentation is to define
the degree to which the technique of the neural networks can
be applied in evaluation of speech deformation in a given type
of pathology. The presented methodology can be directly applied
in the monitoring examinations for patients after larynx oncological
operations and in the future its wider use seems possible in
the evaluation and optimalisation of the results of surgical
treatments in many types of affliction in the larynx, pharynx,
tongue, mouth and paranasal sinusis area.
