While the documentation suggests that the measurements are equivalent to those named in the MDVP, the source code appears to suggest otherwise. The MDVP manual states that they measure "the period-to-period (very short term) variability of the peak-to-peak amplitude". Meanwhile, studying the functions linked below, the Praat shimmer parameters are not "amplitude" shimmer, but rather they are "energy" shimmer (for the luck of a better description).
Do you have a specific reference for this particular way of defining these parameters?
While they are intriguing perturbation measures, they do not measure what most researchers would expect of the parameters which are labeled as shimmer, i.e., amplitude variation (the lineage of Horii 1985). Rather, using the RMS of per-cycle energy they actually reflect both amplitude and period perturbations.
I discovered this oddity recently as I was measuring the shimmers of purely frequency modulated synthesized voice signals. I expected FM signals to have near-zero shimmer measures but yet I was seeing numbers like Shimmer Local = 40%.
While the documentation suggests that the measurements are equivalent to those named in the MDVP, the source code appears to suggest otherwise. The MDVP manual states that they measure "the period-to-period (very short term) variability of the peak-to-peak amplitude". Meanwhile, studying the functions linked below, the Praat shimmer parameters are not "amplitude" shimmer, but rather they are "energy" shimmer (for the luck of a better description).
PointProcess_Sound_getShimmer_local()PointProcess_Sound_to_AmplitudeTier_period()Sound_getHannWindowedRms()Do you have a specific reference for this particular way of defining these parameters?
While they are intriguing perturbation measures, they do not measure what most researchers would expect of the parameters which are labeled as shimmer, i.e., amplitude variation (the lineage of Horii 1985). Rather, using the RMS of per-cycle energy they actually reflect both amplitude and period perturbations.
I discovered this oddity recently as I was measuring the shimmers of purely frequency modulated synthesized voice signals. I expected FM signals to have near-zero shimmer measures but yet I was seeing numbers like Shimmer Local = 40%.