Current system uses linear interpolation.

As to whether it sounds any better, that needs someone with younger and
better ears than mine.

Yes, the cubic interpolation is mainly needed
for wavetable synthesis, and any opcodes that
deal with typical audio sample rates. Opcodes
that read short "simple" waveforms are probably
best left alone (but it doesn't hurt to have variants
which use higher order interpolation, of course), 
because typically these waveforms can be stored at 
whatever sample rate is desired, within reason. That 
is, they can be oversampled a lot, without using up
a lot of memory. The oversampling reduces the interpolation
noise. Wavetable synthesis typically deals
with much longer waveforms, and thus it is usually
not feasible to oversample.

It is great that higher order interpolation is/has been
implemented!

Greg.