Some quick and dirty AC analysis using LTSpice shows some interesting problems:
[Figure 1: Three AC passes using LTspice showing both phase and amplitude. The blend control has been set at low, mid, and high illustrating the change between the diode saturation and hi-pass filter.]
The hi-pass filter has an approximate cutoff frequency of which I'm worried is both ineffectual as a desirable effect, and will always cut some bass so if I wanted to use this to get a really grimy bassline out of a synthesiser it would come out somewhat weak.
To fix this I think the blend control needs more options. I think by adding a three pole switch to the hi-pass filter side that allows selection of either [hi-pass; neutral; lo-pass] settings would be much more user-friendly. This would also be great for the other side so that the user could switch between a couple of sets of diodes.
Back to Figure 1: the blue line shows the characteristic when the blend potentiometer is set mid-way. It shows a drop of around 0.9dB in comparison to when the potentiometer is set to one of the extremes. This is around a Just Noticeable Difference, but for the purpose of this circuit I think it is probably forgivable. It'll have to wait until the breadboard stage to find out if I'm right about this.