If Allow note length for audio samples on the grid is eventually implemented, the current Attack and Decay would still be applied at the beginning and end of the audio event regardless of how its length across steps. Or at least we can specify this assumption there.