Comparing parameterizations of pitch register and its discontinuities at prosodic boundaries for Hungarian

Abstract

We examined how well prosodic boundary strength can be captured by two declination stylization methods as well as by four different representations of pitch register. In the stylization proposed by Liebermann et al. (1985) base-and topline are fitted to peaks and valleys of the pitch contour, whereas in Rei-chel&Mády (2013) these lines are fitted to medians below and above certain pitch percentiles. From each of the stylizations four feature pools were induced representing different aspects of register discontinuity at word boundaries: discontinuities related to the base-, mid-, and topline, as well as to the range between base-and topline. Concerning stylization the medianbased fitting approach turned out to be more robust with respect to declination line crossing errors and yielded base-, topline and range-related discontinuity characteristics with higher correlations to perceived boundary strength. Concerning register representation, for the peak/valley fitting approach the base-and topline patterns showed weaker correspondences to boundary strength than the other feature pools. We furthermore trained generalized linear regression models for boundary strength prediction on each feature pool. It turned out that neither the stylization method nor the register representation had a significant influence on the overall good prediction performance.

Entrainment profiles: Comparison by gender, role, and feature set

“…These lines are subsequently derived by linear regressions, time has been normalized to the range from 0 to 1. As described in further detail in [61] this stylization is less affected by local events as pitch accents and boundary tones and does not need to rely on error-prone detection of local maxima and minima. Based on this stylization the midline is taken as a representation of pitch level.…”

CoPaSul Manual -- Contour-based parametric and superpositional intonation stylization

“…Second, errors resulting from incorrect local peak detection are circumvented. Both enhances stylization robustness as is shown in [21].…”

“…If no tier is specified, syllable boundaries derived by step 7.2 will be selected as candidates. At each boundary candidate a feature set is extracted that had been proven to be related to prosodic boundaries in former studies [20,21]. This feature set is introduced in section 8.9.…”

“…augment:glob:cntr mtd=split augment:glob:prct=mySplitPoint Since for all extracted pause length and pitch discontinuity boundary features are positive correlation has been found to perceived boundary strength [20,21] B and NB centroids can be straight-forwardly derived from high and low feature values, respectively. Centroids are thus derived by splitting each column in the feature matrix at the percentile augment:glob:prct.…”

“…Boundaries are parameterized in terms of discontinuity features of several register representations. Details and an application for perceived prosodic boundary strength prediction can be found in [21].…”