International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 1999)

Florence, Italy
September 1-3, 1999

Neural Networks Techniques for Vocal Fold Pathology Detection

Vyacheslav V. Parshin, Mikhail A. Toumilovich

Belarusian Slate University, Radiophysics Dept., Minsk, Belarus

The digital signal processing techniques are widely used in modern medicine and here there are many areas for investigations directed on enhancement of the effectiveness of these techniques and extension of their possibilities. One of such areas is the development of the automatic means for detection of human speech producing organs pathology based on analysis of their speech. Success in this area would supply the physicians with non-invasive procedure for speech pathology diagnostics that does not causes pain and discomfort to the patient and does not require the subjective evaluation. In this work the problem of speech pathology diagnosis is discussed and is referred as the speech signal classification problem. We suggest using the neural network approach to the problem because it is well developed for many similar problems in speech recognition.

Index Terms. speech processing, neural networks, training techniques, vocal fold pathology

Full Paper

Bibliographic reference.  Parshin, Vyacheslav V. / Toumilovich, Mikhail A. (1999): "Neural networks techniques for vocal fold pathology detection", In MAVEBA-1999, 118-119.