[Joe Bloggs]

Is there anything in the literature that shows that the transition band needs to be widened in this way to shorten the supersonic ringing for anyone who doesn't hear anything at 22kHz? There'll always be some ringing for graph-o-philes to pick at unless you go all the way down to a Q=0.5 lowpass filter, in which case for all I know 192kHz sample rate wouldn't be enough for a flat 20kHz passband :rolleyes:

 

[arnyk]

Is there anything in the literature that shows that the transition band needs to be widened in this way to shorten the supersonic ringing for anyone who doesn't hear anything at 22kHz? There'll always be some ringing for graph-o-philes to pick at unless you go all the way down to a Q=0.5 lowpass filter, in which case for all I know 192kHz sample rate wouldn't be enough for a flat 20kHz passband

 
The existing ringing for the Q=150  filter  (Q= arbitrary quality, not Q in the electrical sense) and 2 KHz Transition Band that goes with it seems to be well within the range of things that seem to be not audibly different from perfection.  It provides good control over aliasing and the like.
 
Questions about very narrow transition bands were raised by this paper, but the paper has some serious issues:
 
https://secure.aes.org/forum/pubs/conventions/?ID=416