The lever is not really shown on my site, but is well described in various AES and SAE papers (see my resume for references). The bottom line is that a horn increases the acoustic coupling as the square root of the mouth to throat area ratio, but a lever does this as a direct ratio. So with an area ratio that would double the output due to the horn, the lever would quadruple it. In reality the two things are hard to compare because they both have completely different sets of tradeoffs and problems.

But I stand by my statement that I do not see the advantage of horns at low frequencies.