I have no idea what your question is. You can either get out what is in the music or you can create something artificial or something in between, so the comment about directivity is a bit moot, especially since the ambience of the hall is captured at some level in the recording already. It you add reflections you may get a sense or space but it is artificial and may be good or bad. If you are sitting in the best seats, i.e. Orchestra, closer to the front, the directed sound dominates though bass will reverb which is not as directional.
50 usec, yes, but again, all that is needed is some level of consistency and it does not even need to be perfect as small differences are equivalent to moving the speaker a bit. Consistency in the crossover region is a bit more difficult depending on the frequency placement, but I am mainly concerned at ITD dominant frequencies assuming ITD data in the music (rare).
in my older post i write more concret, and i enhance it more. My question is. If you think the attack and decay speed of a speaker can not see on step response then just tell me on which measure result can see this better or best ?. all what really hear can measure thats physical fact. In speaker tests often is tell that it have good or bad stereo width. and my tests confirm always fastest step response have best stereo width.
when move the speaker a bit the diffrence stay always the same. maybe ears can correect better. but i notice it give best sound for me when i measure the delay of speaker and correct it in dsp. in carma 4 (freeware) it is very easy and fast to measure delay of speakers. when switch to time and click on right speaker it show the delay of right speaker in samples. delay of left speaker is always set to 0.
look at the waveform i post from dire straits brothers in arms. i now add a box(so lines are exact in angle, so can see better which level it should play on right channel and on left channel at same time. this is the 2 khz lowpass filtered version
there can see on left channel is a part the left speaker must not move much, but the right speaker need move much more. so there is no consistency. I can see on whole waveform of songs that left and right is much diffrent.
if in a speaker the phase can produce exact upto 50 usec period time this speaker is also able to produce frequency of 20 khz. which a slow mid woofer can not doi. t should be clear that a big bass mid woofer that can reach only 6 khz -3 db (0.16 ms period time) have problems to reach the correct level good in period time lower as 0.16 ms which is need for good ITD stereo even if frequency is below 2 khz. because left and right speaker play diffrent signal the time to reach a level variate much more as on a fast speaker.
when hear diffrence in time at lower as 0.05 ms between left and right channel at frequency below 2 khz need a speaker that can produce it correct.