Heh, you can't talk about graphs without showing them, so pls post your measuremetn after correction.
Here are a bunch of curves.
1-Anechoic measurements (third-party)
JBL 305P mkii, measured by Amir :
Neumann KH-120, on axis, measured by kenrockwell.com
Neumann kh-120, sound power, measured by Klippel.de :
We can see that the kh-120 have a very flat on-axis response, but off axis, things get more chaotic.
The 305P shows the opposite behaviour : the on-axis response is not very good, but the off-axis response is faithful to the on-axis one.
Which leads to...
2-pre-equalization of the JBL 305P
Applying this correction (here shown upside down) to the JBL should transform them into a speaker that, on the paper, should have better performances than the KH-120 :
The reason why the correction stops after 4000 Hz is explained in this thread :
https://www.audiosciencereview.com/forum/index.php?threads/jbl-305p-mkii-equalization.10874/
It is because the original measurement was wrong above that frequency, because of reflections on the protection system of the Klippel microphone. The last bump, at 14 kHz, was added later, according to MMM measurements of my own.
3-In room measurements
Here is how these speaker now measure. Mind that the treble switch of the Neumann is set to -1 during all the experiment. Its effect starts at 8 kHz. So it has very little impact on what we see here.
The treble switch of the JBL was at 0, but the bass switch was at -1.5 dB, as far as I remember.
The room has no acoustic treatment. The speakers stand in from of a very large window. The listening point is at 2.1 meters from the speakers, and the room is 6.2 x 3.4 x 2.5 meters.
Both measurements are taken using MMM with pink noise, inside a quite small area (+/- 20 cm around the listening position), then REW Var smoothing was applied.
We can see three behaviours here.
Above 1000 Hz, each speaker has its own frequency response, that depends on its directivity (since the on-axis responses are now supposed to be the same, at least until 8 kHz).
Below 150 Hz, what we see is the effect of the bass reflex port position : in front of the Neumann, but in the back of the JBL. Room modes (54 and 69 Hz) are not fed the same way, nor is the first manifestation of SBIR (100 Hz).
Between these frequencies, the response of both speakers are identical, measured from the listening position.
4-Comparison between sound power and in-room response
These graphs are displayed without smoothing.
We can clearly see the effect of the sound power curves on the final result.
It is difficult to listen to the speakers in this configuration because of the terrible frequency response below 200 Hz.
5 - Room equalization
The room equalization of the Neumann is active from 35 to 800 Hz. It was adjusted frequency by frequency over two years of daily listening. It is now stable and I haven't felt the need to change it for more than one year.
The room eq of the JBL was done in 5 hours, as described in my previous messages. The result shown here is based on the hypothesis that the target curve, according to the directivity index that we saw in the spinorama, should be decreasing from 250 to 2000 Hz.
The level at 2000 Hz was set identical to the level without room eq. The level below 250 Hz was set similar the Neumann's.
The slope was guessed by ear.
Now I can listen to them and compare.
There is a difference in high frequencies. The JBL sound brighter, with a lot of detail. The Neumann sound distant, but natural. As if the musicians were playing farther away.
The female voices sound very natural on the Neumann, but not quite so on the JBL.
The JBL seem to have deeper bass.
6-Clone equalization
Then I tried a different approach. Instead of setting a room eq for each speaker, I tried to equalize the JBL exactly as the Neumann from the listening position, using a 31 bands graphic equalizer.
Here are the measurements :
Though the curves look nearly identical, the voices sound a bit more natural on the Neumann.
In my opinion, it comes from the little bump at 350 Hz. Normally, such small deviations are barely audible, if at all, except if two conditions are present :
-it is the dominant frequency range on the recording
-it is the highest point of the whole frequency response
which are both true in this case.
We can also see that the difference curve is not very flat in high frequencies. My correction was not accurate enough.
7-Conclusion
What I can conclude from this experiment is that even when the speakers have a neutral on axis frequency response, the off axis response has a clearly audible effect on the sound, especially in the high frequencies.
Also, trying to equalize a speaker demands a very high accuracy.
And above all, if the front wall is not treated (here, it is the worst possible case : a window !), it has a very strong effect on the natural of human voices. The equalization of this effect, from 150 to 1000 Hz is possible in theory, but it takes a lot of time and demands an extreme accuracy. Here, this room correction is ok for the Neumann, but after 5 hours of work, I could not get it completely right on the JBL.
Last, here are the same measurements as above, but without smoothing. The effect of the window behind the speakers on the frequency response can be seen as strong vertical variations in the 100 - 500 Hz range. These have certainly an effect on the final result, and equalization can do nothing about it. (mind that the MMM area was a bit smaller than usual : 40 x 40 cm only, which makes the curve look a bit worse than it should).
