I was just thinking about how delaying one input and output by half the delay-time of the samplerate (so 1/(48000*2)sec) could add the possibility of using the stereo input as an oversampled input.
You could then send a single signal both to left and right and read the left and right input serially to get the in-between values that an 48khz samplerate would miss.
Would this in any way be possible?