. . . "This is useful for applications such as voice recognition because it allows setting a higher value for the threshold that triggers recording without missing the initial samples that led up to the first over-threshold sample." .