The usual* sort of error has crept in here.

In calculating the imbalance between the "thick" and "thin" end of the resopnant cavity, the people involved have lost sight of the lateral pressure on the tapered walls of the cavity as it thins.

The nett effect is zero. (think about why the pressure on the base of a tapered glass of water is the same as for the base of a straight glass with the same water level).

Yes, there is an expansive force acting on the cavity, but there shouldn't be any nett imbalanced force.

* When I say "usual" I mean in the usual way of perpetual motion machines and "free energy". There are detailed calculations that obscure the missing logic. If it works, I'll eat my hat**.

** an easy promise, since if it works, we can all ride to the stars and I'll have plenty of time to boil my hat in a good sauce...