Why the music has to be held, not just delayed
A conventional edit would start the drop at a fixed timestamp. This mode holds the song in a loop instead, and releases it when the simulation reaches a state. That difference matters because the ball is what the viewer is watching: if the drop lands at thirty seconds regardless of how big the ball is, the two halves of the video have nothing to do with each other.
Holding also makes the run robust. A ball that gets an unlucky sequence of bounces takes slightly longer to reach its threshold, and the music simply waits. The alternative is a drop that fires while the ball is still visibly small, which is the one failure a viewer will notice immediately.
The practical consequence is that Beat Drops After is a target rather than a hard cut. Set the growth so that the threshold is reached at roughly that time and the two clocks stay in agreement.

