Ummm. I can't really understand your point. The fact is that no loudspeaker is completely omnidirectional - just as well - or the splashing of sound off the side walls (etc.) would sound mighty confusing to the listener at the sweet spot. Let's look at an "old fashioned" polar plot of the type that used to be published years ago.
Originally Posted by STHLS5
Attached is a top-down view of a typical speaker. The speaker is (literally) mounted on a very slowly rotating turntable, taking perhaps a minute or two to turn 360 degrees. The measuring microphone is clamped in the usual position, say 1m from the tweeter at the start position i.e. on the reference axis. The test oscillator outputs one fixed frequency as the speaker completes its 360 degree rotation and as it rotates, the sound pressure generated by the loudspeaker system is collected by the microphone as a continuous voltage. That voltage drives the pen chart's ink pen mechanism which is synchronised with the turntable rotating the speaker box. Once back at 0 degrees, the oscillator frequency is reset and another rotation commences. Thus, we can clearly see how directional the speaker is by the shape of the pressure envelope or balloon.
You can see in this (good) example that at, say, 90 degrees off axis, the output at 10kHz is 15-20dB down compared to say 100Hz. If the frequency was increased to 20kHz, the balloon would be more like a pencil of sound, with almost nothing off axis (drawn in red - my guess). In fact, this is a somewhat simplistic plot and what is really interesting is what happens at the top end of the bass/midrange unit when, by definition it will be significantly beamy (due to its size etc.) yet thanks to the crossover it hands over to the bottom end of the tweeter where it will be far less directional than the bass unit. And subjectively blending that transition, dear readers, is 70% of the entire speaker's design effort in my experience.
Alan A. Shaw
Harbeth Audio UK