Sound & Pitch: To PointProcess (cc)


A command to create a PointProcess from the selected Sound and Pitch objects.
Purpose
to interpret an acoustic periodicity contour as the frequency of an underlying point process (such as the sequence of glottal closures in vocalfold vibration).
Algorithm
The voiced intervals are determined on the basis of the voiced/unvoiced decisions in the Pitch object. For every voiced interval, a number of points (or glottal pulses) is found as follows:

1. The first point t_{1} is the absolute extremum of the amplitude of the Sound, between t_{mid} – T_{0} / 2 and t_{mid} + T_{0} / 2, where t_{mid} is the midpoint of the interval, and T_{0} is the period at t_{mid}, as can be interpolated from the Pitch contour.

2. From this point, we recursively search for points t_{i} to the left until we reach the left edge of the interval. These points must be located between t_{i–1} – 1.2 T_{0}(t_{i–1}) and t_{i–1} – 0.8 T_{0}(t_{i–1}), and the crosscorrelation of the amplitude in its environment [t_{i} – T_{0}(t_{i}) / 2; t_{i} + T_{0}(t_{i}) / 2] with the amplitude of the environment of the existing point t_{i–1} must be maximal (we use parabolic interpolation between samples of the correlation function).

3. The same is done to the right of t_{1}.

4. Though the voiced/unvoiced decision is initially taken by the Pitch contour, points are removed if their correlation value is less than 0.3; furthermore, one extra point may be added at the edge of the voiced interval if its correlation value is greater than 0.7.
Links to this page
© ppgb 19980322