At the moment, I am a bit puzzled, because I thought the whole point of the APS was to appear as Nokia Software to the audio layer... Do you mean that the delay is to big for interactive VoIP (btw: is it possible to quantify that? 200ms? 1000ms?), but for the music player it is not such a big deal if the music is played slightly later?
Would these restrictions also apply for playing audio to the local user only, rather than into the uplink during a call? There are also some applications (like ours, which is concerned with text-to-speech) where this would be just as useful, and where the acceptable delay is likely much longer than for VoIP.
Are use cases like these taken into account in the design of the APS as well?