In a wide-baffle design like the LS1, is there a point at which the baffle is wide enough that the shape of the edges ceases to matter?
It goes without saying that in theory the baffle edges could be far enough from the driver for the sound pressure to have dissipated to the point where the edge diffraction will be too weak to matter. The question is just how wide the baffle would need to be in order for this to be an effective way to suppress the ripple associated with baffle edge diffraction. I don't really have a good sense of how wide the baffle would need to be in order for this to be a highly effective way to suppress diffraction ripple. But if I were to take a guess, my guess would be that with a baffle maybe twice wider than most conventional baffles this effect would be strong enough to be moderately useful, but probably not strong enough to be highly effective at suppressing diffraction ripple.
At the low extreme of a typical tweeter's range, the wavelength is about 5x greater than the diameter of the tweeter diaphragm. The main lobe at this frequency range (for a tweeter) naturally wants to extend well beyond 180 degrees (edge to edge) except that it is blocked by the baffle (assuming there is no wave guide, which changes everything). Since the baffle interferes with the main lobe, sound pressure at the edge of the baffle is appreciable and edge diffraction will be appreciable unless something is done about it. But things are very different at high treble frequencies, because at high treble frequencies, a typical tweeter is so highly directional that the sound pressure at the baffle edge (for a typical rectangular baffle) is too weak to produce significant diffraction ripple.
Now the irony. If edge rounding is modest, leaving the enclosure looking like a box with modestly rounded edges, the effect at suppressing edge diffraction will be limited to high treble frequencies where it isn't a concern anyway. To understand why modest edge rounding doesn't work at the lower treble frequencies where it would be helpful if it actually worked, it is helpful to think about the extent to which the rounding produces a spread in the distance from the tweeter to the edge. In the case of a sharp edge, there is just one distance. When rounding is applied, the edge distance covers a spread. The short end of the spread is the distance from the tweeter to where the rounding begins, and the long end of the spread is the distance from the tweeter to the point where the rounding ends, at the flat side of the enclosure. In order for the edge to not appear very sharp to the wavelength encountering the edge, the spread in the edge distance needs to be at least 1/5 as great as the wavelength, approximately. If the spread in the edge distance is not at least this great, the edge will look very sharp to the wavelength that encounters the edge. Said differently, if the wavelength is five times greater or more (approximately) than the spread in the edge distance, the edge will look very sharp to the wavelength that encounters the edge. Ideally, the spread in the edge distance should cover a full wavelength.
The spread in the edge distance is the same as 1/4 of the circular circumference corresponding to the radius of the roundoff. If you do the simple substitution and simplification, you find that the wavelength where the roundoff just begins to be a little bit effective is where the wavelength is 8x greater than the radius of the roundoff. For example, if the radius of the roundoff is 1/2", it will start to be a little bit effective at wavelength equal to 4", which corresponds to about 3.4 kHz. It becomes increasingly effective as frequency moves higher, and at lower frequency, the spread in the edge distance isn't great enough to be effective. This frequency is not a whole lot lower than the frequency where a typical tweeter becomes too highly directional to illuminate the edge of the baffle. The gist of it is that edge rounding is of no particular benefit unless the rounding is done to the extent that the enclosure takes on the appearance of a cylinder or a cone. Cylinders and cones are very effective at suppressing diffraction ripple. A bevel is very effective if the bevel clips the edges to the extent where the bevels are as prominent as the front of the baffle.
The most effective way to suppress baffle edge diffraction, for a speaker enclosure that looks more like a rectangular box than a cylinder or cone, is to use a modest waveguide. The modest waveguide has a strong affect on the tweeter directivity at the low end of the tweeter's range, preventing the tweeter from illuminating the baffle edges. At the high end of the tweeter's range the directivity is naturally such that the waveguide has little if any effect at further increase in directivity. A compact waveguide such as the DXT can be combined with strong beveling, and this combination should be highly effective at suppressing edge diffraction. Another benefit of this compact waveguide is that it permits the tweeter to be located closer to the midrange (or mid-woofer), which is beneficial to the polar response in the vertical plane.