Frame-Based Acoustic Cues of Vocal Dysperiodicity in Connected Speech

Acoustic analysis of connected speech is carried out by means of a generalized variogram to extract vocal dysperiocities. A segmental signal-to-dysperiodicity ratio (SDRSEG) is used to predict the perceived degree of hoarseness. The corpora comprise four French sentences as well as vowels [a] produced by 22 male and female normophonic and dysphonic speakers. It is shown that the average SDRSEG correlates better with perceptual scores of hoarseness than the global signal-to-dysperiodicity ratio (SDR). The perceptual scores are based on comparative judgments by six listeners of pairs of speech samples. In addition, multiple linear regression analysis of four statistical descriptors of SDRSEG is performed to gauge whether higher-order statistics are valid predictors of perceived hoarseness