Dialog Behavior: How Sampling Intervals Affect Estimates of Sound and Silence

Abstract
In using computerized systems for analyzing temporal patterns of sounds and silence arising from two-speaker dialog, sampling rates alter estimates of durations of these vocal parameters. Results are reported using the same vocal samples, altering only the sampling intervals, from 2 to 100 samples per second. Significant differences were found in mean durations of pauses and vocalizations among the sampling intervals. An optimal sampling interval of 10 samples per second is suggested.