Yeah if inspecting single reflection only it is quite easy to check out what is needed to get the resulting interference less. About 10db attenuation towards the reflection is needed (well, the reflected signal needs to be attenuated, could be room treatment as well) to get nulls "only ~6db". 20db attenuation makes the interference to ~+/-1db which I think starts to be inaudible. This is the goal I think, but very hard to achieve unless very narrow pattern.
View attachment 1048920
View attachment 1048919
View attachment 1048918
View attachment 1048917
View attachment 1048916
The response can be optimized (like supercardioid) so we can have -20db or more attenuation towards some angles, towards the reflection point, but it is hard to get wide bandwidth, notch is easy. IF we only have a notch like attenuation it doesn't do much, or perhaps even makes the situation worse (speculating that hearing system can mostly ignore the interference if it resembles the direct sound).
Here is reflected signal with EQ dip strategically positioned to a null. It kills tip of the null but right next to it the interference stays as the attenuation wears off. We gotta remember that next to a null is constructive interference as well!
View attachment 1048926
Perhaps any attenuation is good and all we really have to do is to help the brain to filter the reflections out from perception. Interesting question is how we achieve this, how to help the brain? Perhaps attenuation is not as important (so that any attenuation is fine) as keeping the reflection same as direct sound?
We have to remember the reflections don't happen alone but there are plenty which all combine. If one makes similar test but with several reflections it is clearly seen that the reflections "overpower" listening position response over the direct sound. Direct sound is just one signal, and if there is 6 first reflections from room boundaries with different frequency response the resulting response looks more like the reflections and not the direct sound. What basically happens we arrived to the power response. Anyway, it is interesting excercise to figure out if some of these reflections are bad for perceived sound and how to make them less bad, even though the power response was nice. Perhaps try make smooth ERDI to prioritize first reflections over other directions.