Abstract
Nonlinear phenomena such as subharmonics, biphonation, or deterministic chaos induce characteristic patterns in pitch and amplitude contours. A novel measure of correlations in contours termed MAC (mean average correlation) is introduced. For uncorrelated and weakly correlated noise, the expectation values and standard deviations of MAC are calculated. It is shown that trend elimination induces spurious correlations. Moreover, simulations of the circle map are analyzed and high values of MAC for subharmonic and chaotic data are found. Examples are presented of severely rough voices and excised larynx data and correlations are quantified in the corresponding contours by MAC. Finally, the connection between perceived voice quality and MAC is studied with the aid of 120 perceptually evaluated voices. [Work supported by the Deutsche Forschungsgemeinschaft.]
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
