@Clint apologies for the delay, the STL is here.
This version is a little easier to remove and is probably slightly better about pooling moisture. At some point I'll probably do a V3 that's a little taller--this one is still hard to grab with my fingers, and I think it is probably best to be able to take it out to let it dry out.
@Clint though it's hard to see the exact shape at the bell end, I believe there is a continuous taper from the mouthpiece diameter to the output port diameter. So the idea was to create a ring of tabs that bow slightly outward. This gives just enough flex to get it in without much force, but keep it snugly in presumably without putting much deformation pressure on the tube inside due to the taper of the tube.
Fundamentally it's just three stacked rings: an outer 8.5mm diameter ring to keep it from migrating up the tube, the "plug" itself at 6mm, and a smaller ring on top of that to support the tabs and reduce the risk that they snap from too much flex. The tabs bow out 0.25mm, which is plenty to provide a snug fit. I can easily push it in by hand, but need pliers to pull it out because it's too small to grab with my fingers.
One hopefully minor issue is that water can and does pool in between the tabs; going to have to keep an eye on that. I might try a version without the middle tier, though I suspect it will pass more air. Would like to refine a bit more but happy to share the STL.
Yesterday I prototyped and 3d printed a small plug that effectively reduces the output port diameter to about 2.75mm (vs 6mm out-of-the-box), which feels pretty good to me. It does increase the breath noise of the instrument a bit as the air is forced through the smaller opening, but with headphones I don't notice, and overall it seems much more consistent than the piece of tape I started with.
@Support_AODYO is there guidance on how much constriction is too much constriction?
Are there any plans to support high-resolution MIDI control messages for breath? Particularly at the low end, I'm finding the quantization to 7-bit MIDI signals to be very noticeable compared to the built-in sounds, so I assume the breath sensor is more sensitive than the 7-bit range represented by a single control message. Would be great to have access to that sensitivity over MIDI as well!