Abstract
The identification of voice pathologies using speech processing techniques can be an useful contribution to the diagnose of larynx diseases. This work presents an identification system using spectral features and pitch jitter extracted from the sustained vowels /a/, /e/ and /i/. A SVM (Support Vector Machine) classifier has been implemented to discriminate Reinke's edema pathologies from nodules. The main objective of this work is to inspect if the spectral features reported in previous work for the vowel /a/ are present in others vowels, and evaluate the voice pathologies identification rate.
Original language | English |
---|---|
Title of host publication | CONFERENCE ON ELECTRONICS, TELECOMMUNICATIONS AND COMPUTERS - CETC 2013 |
Editors | Alessandro Fantoni, Artur J. Ferreira |
Place of Publication | Amsterdam |
Publisher | ELSEVIER SCIENCE BV |
Pages | 202–208 |
DOIs | |
Publication status | Published - 2014 |
Event | 2nd Conference on Electronics, Telecommunications, and Computers (CETC) - Lisbon, Portugal Duration: 5 Dec 2013 → 6 Dec 2013 Conference number: 2nd |
Publication series
Name | Procedia Technology |
---|---|
Publisher | ELSEVIER SCIENCE BV |
Volume | 17 |
ISSN (Print) | 2212-0173 |
Conference
Conference | 2nd Conference on Electronics, Telecommunications, and Computers (CETC) |
---|---|
Abbreviated title | CETC |
Country/Territory | Portugal |
City | Lisbon |
Period | 5/12/13 → 6/12/13 |
Keywords
- Voice pathologies identification
- formant frequency
- formant bandwidth
- pitch jitter
- pitch