mattbolton
New Member
- Joined
- Aug 28, 2026
- Messages
- 1
- Likes
- 9
Hi Everyone!
Matt here, ive been an avid reader of this forum for many years, however thought its about time I started sharing some of the things im working on. Specifically this is in regards to a web based horn design and directivity simulator. I’d love any feedback, critical or otherwise, and of course feel free to use if you think its useful.
It’s a tool specifically aimed at creating an intuitive, user friendly web based approach to front facing horn and waveguide design. (also tentatively multi entry horns). What ive particularly worked hard on, and what i think makes this a little bit different from what else is out there, is the novel way it tackles directivity modelling.
You can see the tool here:
https://horn-tool.bobohifi.com.au/
Before anything else, credit where it's due: Hornresp has been the backbone of DIY horn design for a long time, and the BEM work coming out of JWSound (Boundary Lab, Polar Generator) is genuinely impressive. This isn't meant to replace either of those. I put this together because I'm a pretty visual person and wanted something a bit friendlier to actually look at and play with while designing, and because I was curious whether there was a way to get a more honest picture of directivity without needing to run a full BEM sim every time I wanted to test an idea.
And just to be clear from the get go, yes, AI was used to help build this. I'm reasonably technical, but I'm not a programmer, and I've used Claude to help put it together. My opinion of claude is that its really good at following direction and putting a working web tool together quickly, but it needs a fair bit of steering to land on genuinely novel solutions. Its also great at doing the legwork to turn an idea into working code.
And now to the tool…
Now generally the horn tool allows for a visual approach to building a multi segmented horn, and results in the electro acoustic response calculations that we find in other tools. Its set up in a way that should allow for tweaking of the horn design/shape and you immediately can see the result in 3d, and in the response curve. It also lays out the information so that the horn shape can be constructed (segment lengths and profiles).
Where most of my time was spent was work on two methods to model directivity and output directivity heat graphs, each in browser.
The Simple Approach. Modes Hybrid.
In the main part of the horn tool, its a step further than a simple diffraction model and an attempt to track higher-order modes building up through the horn's interior as the cross-section changes. This is done with a plane-wave model inside the horn, and a diffraction integral over the horns real aperture shape at the mouth. It also integrates a curved wavefront correction. It's not full BEM, but it's a step closer than a mouth-only model, and it's instant instead of an overnight run. It seems to respond properly to multi-segment horns. Flare law, segment length and interior shape all feed into the polar map. Although at a very quick, rough resolution. Enough to give someone an idea on how different horn profiles behave.
The Hard Approach. FDTD Solver.
Ive also worked on a FDTD (Finite-Difference Time-Domain) solver. This can be found as a “Open in FDTD Solver) button below the polar maps in the horn tool. Instead of the dense computing power required for BEM (progressive boundary calculation for each frequency), FDTD divides the air itself into a 3D grid and steps the wave equation forward in time across it. So one broadband run gives you the whole frequency range at once, and the arithmetic per cell is simple and identical everywhere. This calculation is handled quite well by modern day graphics cards (myself using apple silicon). Now I will note, it does take some time do do a sweep, so ive built in resolution controls, and the ability to reduce the bounds of the calculation.
Generally both models seem to work and correlate to each other. However i am looking for feedback as i dont have many examples of directivity maps corresponding to horns with real dimensions that I can check against.
Anyway, id really appreciate any feedback from people here who've build horns, and especially any validation of the graphs against measurements or BEM results if you have them. Happy to answer any questions if anyone has any!
Cheers,
Matt
Matt here, ive been an avid reader of this forum for many years, however thought its about time I started sharing some of the things im working on. Specifically this is in regards to a web based horn design and directivity simulator. I’d love any feedback, critical or otherwise, and of course feel free to use if you think its useful.
It’s a tool specifically aimed at creating an intuitive, user friendly web based approach to front facing horn and waveguide design. (also tentatively multi entry horns). What ive particularly worked hard on, and what i think makes this a little bit different from what else is out there, is the novel way it tackles directivity modelling.
You can see the tool here:
https://horn-tool.bobohifi.com.au/
Before anything else, credit where it's due: Hornresp has been the backbone of DIY horn design for a long time, and the BEM work coming out of JWSound (Boundary Lab, Polar Generator) is genuinely impressive. This isn't meant to replace either of those. I put this together because I'm a pretty visual person and wanted something a bit friendlier to actually look at and play with while designing, and because I was curious whether there was a way to get a more honest picture of directivity without needing to run a full BEM sim every time I wanted to test an idea.
And just to be clear from the get go, yes, AI was used to help build this. I'm reasonably technical, but I'm not a programmer, and I've used Claude to help put it together. My opinion of claude is that its really good at following direction and putting a working web tool together quickly, but it needs a fair bit of steering to land on genuinely novel solutions. Its also great at doing the legwork to turn an idea into working code.
And now to the tool…
Now generally the horn tool allows for a visual approach to building a multi segmented horn, and results in the electro acoustic response calculations that we find in other tools. Its set up in a way that should allow for tweaking of the horn design/shape and you immediately can see the result in 3d, and in the response curve. It also lays out the information so that the horn shape can be constructed (segment lengths and profiles).
Where most of my time was spent was work on two methods to model directivity and output directivity heat graphs, each in browser.
The Simple Approach. Modes Hybrid.
In the main part of the horn tool, its a step further than a simple diffraction model and an attempt to track higher-order modes building up through the horn's interior as the cross-section changes. This is done with a plane-wave model inside the horn, and a diffraction integral over the horns real aperture shape at the mouth. It also integrates a curved wavefront correction. It's not full BEM, but it's a step closer than a mouth-only model, and it's instant instead of an overnight run. It seems to respond properly to multi-segment horns. Flare law, segment length and interior shape all feed into the polar map. Although at a very quick, rough resolution. Enough to give someone an idea on how different horn profiles behave.
The Hard Approach. FDTD Solver.
Ive also worked on a FDTD (Finite-Difference Time-Domain) solver. This can be found as a “Open in FDTD Solver) button below the polar maps in the horn tool. Instead of the dense computing power required for BEM (progressive boundary calculation for each frequency), FDTD divides the air itself into a 3D grid and steps the wave equation forward in time across it. So one broadband run gives you the whole frequency range at once, and the arithmetic per cell is simple and identical everywhere. This calculation is handled quite well by modern day graphics cards (myself using apple silicon). Now I will note, it does take some time do do a sweep, so ive built in resolution controls, and the ability to reduce the bounds of the calculation.
Generally both models seem to work and correlate to each other. However i am looking for feedback as i dont have many examples of directivity maps corresponding to horns with real dimensions that I can check against.
Anyway, id really appreciate any feedback from people here who've build horns, and especially any validation of the graphs against measurements or BEM results if you have them. Happy to answer any questions if anyone has any!
Cheers,
Matt
