Abstract:
In this paper, vowel sounds produced by healthy speakers and those produced by speakers suffering from certain pathologies are compared using differential trans-spectrograms, or the ``Caseygrams.'' These Caseygrams are generated as follows: First, the vowel sound samples are optimally reconstructed to reduce leakage errors. The data are then partitioned into many smaller samples. Each of these is transformed into the frequency domain. These FFTs are used to calculate two different types of trans-spectral coherences (TSCs). [Please see Vaidya and Anderson, J. Acoust. Soc. Am. 89, 2370--2378 (1991).] These two TSCs are specifically developed to measure the interaction of the main harmonics with the subharmonics. Positive results in this differential trans-spectrogram (Caseygram) indicate that chaos is present, either in a nascent state or fully developed state. It is expected that these Caseygrams would give an early warning of impending disorders and would also help monitor recovery from disorders such as damage to the speech center of the brain following a stroke.