Since no one seem to have an answer, I would suggest just try it with a few audio track with varying lengths and see what the result might be. There are two places where it could be used; one is on compiling sonf tracks and the other is when doing automatic track recognition. That's just my guess.