S.A. Bentil and Y. Yan (USA)
Vocal quality, voice characterization wavelet transform, spectrogram
We present a method to quantitatively characterize voice abnormalities using wavelet analysis as a preprocessing step. The proposed method uses wavelets to decompose acoustic and electroglottography (EGG) signals, acquired clinically from patients with normal and pathological voices, into their designated frequency/time components. These components contain valuable information on the unique dynamic properties of the vocal system and we can correlate these properties with specific voice conditions. A comparative analysis of the vocal signals acquired from a patient suffering from muscular tension dysphonia (MTD) before and after treatment using spectrogram and FFT is also presented. These combined analyses provide a comprehensive representation and quantitative characteristics of the vocal dynamics, which may provide key indication of vocal abnormalities.
Important Links:
Go Back
