东华大学学报(英文版)
東華大學學報(英文版)
동화대학학보(영문판)
JOURNAL OF DONG HUA UNIVERSITY(ENGLISH EDITION)
2011年
1期
5-9
,共5页
vocal cord paralysis%wavelet packet analysis ( WPA )%support vector machine (SVM)%anaesthesia
Vocal cord paralysis can occur as a complication of surgery or anaesthesia, if permanent is a significant clinical problem. Early detection is important to optimize the chance of repair, and avoid complications associated with an impaired swallow. An algorithm to detect altered vocal cord function was presented based on wavelet packet analysis ( WPA) and support vector machines ( SVM ) , and compared with the Hoarseness Diagram method ( HDm), which was reported as an objective voice quality evaluation approach and could be used for pathological voice discrimination. Experiments using voice signals recorded from subjects before and after the procedure show high classification accuracy with the new algorithm, whereas HDm fails in the detection of a hoarse voice. This finding would help to develop a screening tool to detect the vocal structure damage during surgery.