| Abstract | Objective: This study aimed to investigate the value of using multiple vocal acoustic classifiers representing different aspects of the disordered vocal signals such as glottal noise, periodicity, signal stability, spectral slope, and overall voice quality in discriminating voice-disordered female speakers from non-disordered female speakers using a vowel task.
Methods: Vowel /ɑ/ samples were extracted from 133 voice-disordered female patients and 97 non-voice disordered female speakers and were signal typed prior to analysis. Praat software was used to measure Harmonics-to-Noise ratio (HNR), Glottal-to-Noise Excitation Ratio (GNE), the standard deviation of fundamental frequency (F0SD), and Cepstral Peak Prominence (CPPp); and the Analysis of Dysphonia in Speech and Voice (ADSV) program was used to measure CPPadsv, Low/High spectral ratio (LH), and the Cepstral/Spectral Index of Dysphonia (CSID). Outcome measures included sensitivity, specificity, and classification accuracy.
Results: As single acoustic measures, only ADSV-based measures showed good (CPPadsv) and acceptable (CSID) discrimination results. For the Praat-based measures, HNR, GNE, and CPPp had acceptable sensitivity but poor or non-acceptable specificity and classification accuracy. Logistic regression models with all Praat measures (F0SD, HNR, GNE) plus ADSV measures (CPPadsv, LH, or CSID) provided excellent sensitivity, good-to-excellent specificity, and excellent classification accuracy. ROC analysis showed that CPPadsv, CSID, CPPp, GNE, and F0SD had the highest AUC values.
Conclusion: A combination of acoustic classifiers representing major aspects of vocal dysfunction resulted in good to excellent voice classification outcomes. Single acoustic measures had lower classification ability than combined measures. The findings implied that acoustic measures extracted from a vowel were useful in voice disorder classification.
|
|---|