The goal of the frustrum is to thin out the horn wall where the hole goes through while minimizing the amount of air trapped between the speaker cone and the horn wall. You've done a nice job of that but there is more to it and that. The Danley patent shows an example where there is no tapering and that can work well too. It all depends....on a lot of different things. You really need to do a simulation (e.g. HornResp) where you tune hole diameter, wall thickness, trapped air volume, hole location, .... to get the best response.