Thanks amir! That 1Kish bump is really weird given it's not in Harman's data but I have no good explanation for it =] Just seems like they definitely would've EQ'd out such a broad hump in DSP so I can't help but suspect something odd is going on. And again we see the shelving down in the bass...
While I understand not EQing out the small ripples, such a broad hump would absolutely be audible and it's hard to believe it would have passed the testing phase that way when it seems trivial to EQ it out since the DI curves are quite smooth in the region.
To be fair to JBL, they do quote the frequency range as being to the F10 point in their detailed spec sheets (
PDF), and they generally specify it as F10 in all of the specs I've seen.
Also a gentle reminder against putting
too much weight on the preference score, especially for speakers so close in predicted performance. Remember, accuracy is pegged at 86 percent, but that still leaves a whole lot of room for variation and outliers. And ultimately the score is referenced to listening, not the other way around. Granted, blind listening, but still.
To use an example from the preference study itself:
View attachment 53531
One speaker was predicted at roughly a 4.5, another at a 5, so pretty close to the 705P and the Kali. In actuality, the
measured performance of the former was a measly 2, while the other was nearly a 7. You can also see there's another speaker that was predicted at a 5 but in actuality scored roughly a 3.5. There's a whole lot of wiggle room there so it's really just to numerically compare speakers that perform really poorly and really well...