No. I developed equalization for what the frequency response measurements showed to be in error. The correction was barely audible and not at all how it looks on the graph. Our auditory bandwidth is enlarging as frequencies go up which reduce the impact of such aberrations. In addition, you have to have music content that hits on those specific notes, and know in advance that such notes should sound different to detect it as something being wrong. This is asking a lot from a tester. Harman's research showed their trained listeners could detect response variations in 6 bands. That is heck of a lot wider than this very narrow response error.
If you're trying to convince me that you seeing the measurements before hand doesn't bias you in any way, then we'll have to agree to disagree on that
. Also, it seems you're getting a bit defensive towards something that wasn't meant as an attack against you. Probably poorly stated on my part, but I'm actually praising you here for listening first. I really hope it continues, even if it's just time to time. I don't think it's possible for our brain to not be biased to at least some degree. I recognize the advantages of measuring first, but I think the advantages of listening first outweigh those. Ultimately, what you did here (listen > measure > EQ > listen) is the best of both worlds. Thank you for taking the time to do it.
Take that error out and the rest of the response is quite smooth,
I'm not sure I agree with this. This a 4dB peak resonance from 800-1,100Hz. Assuming that's the error you're talking about? The cancellation on its own(seen best in the DI) is not what I'm talking about here.
For one, I don't think it's fair to say ~"so remove a 4dB positive magnitude 300-400Hz wide resonance" right at one of the most critical sections (800-1100Hz). I think a lot of speakers could look quite a bit smoother if you remove their biggest flaw.
Second, calling the remaining response "quite smooth" I still wouldn't agree with. You still have a big hump at 110Hz, a 2dB magnitude fairly wide depression from 200-500/600Hz, then a resonance at 1.5kHz, followed by a dip at 1.8-2kHz, then a resonance at 2.5kHz, then a 1-2dB rise from 2.8-4kHz, then a 3 to 4dB drop from 4kHz to 10kHz, then a resonance at 11kHz, and finally another one at 15kHz. To make things easier to see, I've put a black box over the 400Hz wide section you say to "take out".
The remaining response is still not what I would call "quite smooth", for a $2,000 speaker. I would describe it more as "rough, but not bad"(overall shape is still decent). Could just be a difference of subjective opinion, though. Here is a side by side comparison with a $79 speaker that you gave a 0/5, and described as "fairly uneven" and with "multiple resonances".
That's a $2,000 speaker you rated 5/5 against a $79 speaker you rated 0/5. The commentary is also very different. My point is, subjective impression rules the day when it comes to panther score. Frequency response, price, etc. are obviously a much smaller factor. Like I said, I like it this way. We essentially have both a subjective(panther) and objective(Olive) score for every speaker.
combined with directivity that is perfect meaning reflections are in very good sync with on-axis.
This I agree with 100%, and was I point I brought up in my first or second post. This is up there with the best directivity we've ever seen. To me, it looks very similar to the DIYSG HTM-12 v2 that
@hardisj measured(which also has several resonances).
I think trying to come up with a new modified Olive score that matches even closer would be a fun task for this site. Maybe weighing directivity more is one possible way to improve it? This thing has quite a few resonances, but with the SOTA directivity it has, those resonances were clearly not enough of a factor to hurt it.
None of this combines to say the measurements are "poor." They aren't. We have one anomaly of note and I determined through careful, blind, controlled testing that it only had a subtle effect.
Maybe we have different ideas of what "poor" measurements constitute. I would describe the above graph on the left, for a $2,000 speaker, as "poor". I also don't agree with judging something after removing "one anomaly", when that 1 anomaly is a 300Hz wide 4dB positive amplitude resonance centered at ~900Hz.
Now it is possible I am wrong. But it is just as well that your read of these measurements is wrong.
It's definitely possible I'm wrong, and that's part of why I'm responding. I'm hoping these side by side examples will help to give context to others. They can see whether or not my assessment of this speakers frequency response as "poor" is being too harsh. Not saying terrible, but "poor" is just the best word I could think of to describe what I see. Keep in mind I'm talking strictly about the FR here. As mentioned, directivity is excellent. I also think
@napilopez makes a great point about the overall shape of the curve being good. Despite the many resonances, I bet this would look "quite smooth" with good smoothing.
This review should make people think. It made me think as to why the response error was not audible so I ran experiments to prove that to myself. You need to do the same. Don't just pontificate about what I did or did not do.
Agreed. I think this speaker would be an excellent choice to involve in a blind listening comparison. It's got quite a few resonances and response errors, but the overall slope is decent, and the directivity is SOTA. I'll keep my eye out for a used set
.