In the Xenos 0HA-REP, most probably, there is something of a bass cut if you bypass the output buffers. That's due to those 100uF 16V electrolytics going to ground on the feedback loop of the gain op-amp (that is, where you put the AD8397). I'm not 100% sure, but most probably if that op-amp is made to drive a low impedance their capacity will form a high pass filter.
On the other hand, in this particular design those two caps are needed to take care of the DC offset with bipolar opamps like the NE5532 (and therefore the AD8397). Jumpering them, I noticed a definite increase of DC offset even with the LM4562 (and output buffers). Of course no bass cut for sure if you keep the buffers.
Btw, here's a hint -- try the AD811 for the output buffers. I just discovered these and they look just *perfect*, especially if you can deal with the fact that they're only specified for +/-15V and +/-5V, and that their quiescent current is a conspicuous 14.5 mA each.