A command that becomes available in the Query submenu when you select a PointProcess object.
This command will write into the Info window the Difference of Differences of Periods, a jitter measure defined as the average absolute difference between the consecutives differences between consecutive intervals, divided by the average interval (an interval is the time between two consecutive points).
As jitter is often used as a measure of voice quality (see Voice 2. Jitter), the intervals are often considered to be glottal periods. For this reason, the command has settings that can limit the possible duration of the interval (or period) or the possible difference in the durations of consecutive intervals (periods).
The jitter can be used as a measure of voice quality. See Voice 2. Jitter.
(In the following the term absolute means two different things: (1) the absolute (i.e. non-negative) value of a real number, and (2) the opposite of relative.)
DDP is defined as the relative mean absolute (i.e. non-negative) third-order difference of the point process (= the second-order difference of the interval process), as follows.
First, we define the absolute (i.e. non-relative) Average Perturbation (in seconds) as one third of the mean absolute (non-negative) difference of difference of consecutive intervals:
|absDDP(seconds) = ∑i=2N-1 |(Ti+1 - Ti) - (Ti - Ti-1)| / (N - 2)|
where Ti is the duration of the ith interval and N is the number of intervals. If an interval Ti-1 or Ti or Ti+1 is not between Period floor and Period ceiling, or if Ti-1/Ti or Ti/Ti-1 or Ti+1/Ti or Ti/Ti+1 is greater than Maximum period factor, the term |2Ti - Ti-1 - Ti+1| is not counted in the sum, and N is lowered by 1 (if N ends up being less than 3, the result of the command is undefined).
Second, we define the mean period as
|meanPeriod(seconds) = ∑i=1N Ti / N|
where Ti is the duration of the ith interval and N is the number of intervals. If an interval Ti is not between Period floor and Period ceiling, or if Ti-1/Ti or Ti/Ti-1 is greater than Maximum period factor and Ti+1/Ti or Ti/Ti+1 is greater than Maximum period factor, the term Ti is not counted in the sum, and N is lowered by 1; this procedure ensures that in the computation of the mean period we use at least all the intervals that had taken part in the computation of DDP.
Finally, we compute DDP as
|DDP = absDDP(seconds) / meanPeriod(seconds)|
The result is exactly 3 times the RAP jitter measurement: a value between 0 and 6, or between 0 and 600 percent.
© ppgb, March 2, 2011