KEF R3 Speaker Review

[fmplayer]

IMO, point is that actual standard targets for speakers are only "static" in that they don't objectivate the dynamic behaviour. We don't know for sure yet if a subjective response curve wouldn't be the translation of a particular frequency dependent variation of the dynamic behaviour. Or if this correlates with a somewhat lifeless character, as quoted before.

One would be very surprised how a multiband compressor can make a bass shy record sound right and impactful just by introducing some specific bass compression (repeat compression) and a delicate increase of bass (please read Bob Katz's 'Mastering Audio' to convince yourself). Or by the influence of a master bus compressor to the sound (often used during mastering process).
You can immediately see the use of the latter on the waveform of a music file.
Long story short, there are some measurements lacking to correlate what we hear with what is measured.

And that is not yet (to my knowledge) covered by measurements, so we need to find additional measurements.

 

[Chromatischism]

IMO, point is that actual standard targets for speakers are only "static" in that they don't objectivate the dynamic behaviour. We don't know for sure yet if a subjective response curve wouldn't be the translation of a particular frequency dependent variation of the dynamic behaviour. Or if this correlates with a somewhat lifeless character, as quoted before.

One would be very surprised how a multiband compressor can make a bass shy record sound right and impactful just by introducing some specific bass compression (repeat compression) and a delicate increase of bass (please read Bob Katz's 'Mastering Audio' to convince yourself). Or by the influence of a master bus compressor to the sound (often used during mastering process).
You can immediately see the use of the latter on the waveform of a music file.
Long story short, there are some measurements lacking to correlate what we hear with what is measured.

And that is not yet (to my knowledge) covered by measurements, so we need to find additional measurements.

I think this is all the job of content. If your content doesn't move people, make different content. Speakers should simply reproduce content as faithfully as possible. That means designing speakers that come close to the research done over the last several decades by Toole, Olive, Welti, Devantier, and more and are graded and reviewed here at ASR.

 

[fmplayer]

I think this is all the job of content. If your content doesn't move people, make different content. Speakers should simply reproduce content as faithfully as possible. That means designing speakers that come close to the research done over the last several decades by Toole, Olive, Welti, Devantier, and more and are graded and reviewed here at ASR.

Indeed, but how do you assess faithfulness ? Either you compare by listening a track on your system with the same played on the system which served for mastering. Or you have a set of measurements that objectivate it.

We all thank @amirm for offering this excellent site to our community. I admire and thank the people you quoted for all the science they brought to our subject of interest. Applying their findings is with no doubt necessary. And, in order to stick with the thread, R3s follow undoubtedly the findings.
But is it sufficient ?
In other words: are we sure we know everything (or at least the most part) that has to be known for certainty? If the answer is yes, then
a. listening should not be needed to have an opinion about loudspeakers
b. speakers that check all the boxes could sound only very good. (really ?)
c. there would be no need to further investigate.

For what I'm aware of, research is still actively on the agenda, and slowly but surely our knowledge improves. When it will be considered sufficient by the stakeholders, research will switch to other points of interest.

 

[Sancus]

I speculated that they failed the subjective listening test

They failed it due to a room mode, this was specified in a subsequent review. Unfortunately it keeps coming up because a correction was never added to this review. This is at least the 5th time I've seen it come up. Perhaps @amirm would consider adding the below quote to this review's listening test section:

Focusing below 200 Hz, we see our loudest and most offending peak is around 100 Hz. Let's dial in a single parametric filter by eye and see what does[...] Beauty of Roon is that its filters can be switched on and off in a second. Boy, was that a miracle fix! Gorgeous detail was there with almost no loss in total bass energy. Indeed bass was now tighter. The "magic" that I heard in the Revel M16 was now imparted into KEF R3.

 

[richard12511]

I speculated that they failed the subjective listening test because of intermodulation distortion, which is a problem known to plague coaxial drivers. It wasn't measured in the review.

I tend to believe that whatever we can hear in a speaker can be measured as well. Subjectivists will say that measurements don't explain everything. This is true if you have a limited set of measurements. However, the solution is more measurements to get a better understanding, not chalking up the disparity between what we hear and a limited set of measurements to magic or hidden variables.

Really doubt it was due to the IMD, given the 3 way design. I think Amir is used to the Revel wide dispersion sound as his reference, and he tends to judge speakers based on how close they get to that. Directivity is almost certainly the main culprit here, imo. I also think mono listening tends to highlight wider dispersion more than stereo listening does, though that's an opinion not well supported by science(yet ). Second most likely cause is probably that low Q mid/upper mid dip followed by the low Q presence and brilliance region peak. I guess we don't know for sure, though since we don't have IMD measurements here.

 

[richard12511]

They failed it due to a room mode, this was specified in a subsequent review. Unfortunately it keeps coming up because a correction was never added to this review. This is at least the 5th time I've seen it come up. Perhaps @amirm would consider adding the below quote to this review's listening test section:

Yeah, that definitely needs to be updated, and he needs to photoshop a golfing panther over the one that's currently there. Kinda unfair to KEF not to. Perhaps someone can help with the photoshop.

 

[nerdoldnerdith]

They failed it due to a room mode, this was specified in a subsequent review. Unfortunately it keeps coming up because a correction was never added to this review. This is at least the 5th time I've seen it come up. Perhaps @amirm would consider adding the below quote to this review's listening test section:

Ah. Well, I didn't read all 50 pages of this thread, so I guess I missed that. For what it's worth, I did delete my original comment when I searched the thread to find that my point had already been made. Alas, everything you post on the internet lasts forever...

Even considering the revelation, it would still be nice to see IMD measurements, especially for speakers with coaxial drivers like the R3.

 

[Chromatischism]

Indeed, but how do you assess faithfulness ? Either you compare by listening a track on your system with the same played on the system which served for mastering. Or you have a set of measurements that objectivate it.

Easy. You measure the speaker's linearity. If it plays back all frequencies at the same SPL over your listening area, it will play back recordings faithfully. Considering the challenges in designing speakers that behave off-axis as well as on-axis, this is easier said than done. But not impossible as we have some good examples reviewed here so far on this site.

Note, I do not mean that all speakers will sound the same. There are small differences in some measurements that actually affect what we hear a great deal. So let's say faithfulness within +/- a few % which is pretty good. Whatever number you think is within the range of neutrality.

In other words: are we sure we know everything (or at least the most part) that has to be known for certainty? If the answer is yes, then
a. listening should not be needed to have an opinion about loudspeakers
b. speakers that check all the boxes could sound only very good. (really ?)
c. there would be no need to further investigate.

For what I'm aware of, research is still actively on the agenda, and slowly but surely our knowledge improves. When it will be considered sufficient by the stakeholders, research will switch to other points of interest.

Right, we don't know everything and I don't think anyone would claim that. If that were true, we'd all pack up and go home. There would be no more science left to be done.

 

[fmplayer]

Easy. You measure the speaker's linearity. If it plays back all frequencies at the same SPL over your listening area, it will play back recordings faithfully.

That's my point. Is it stated (and proved) somewhere by somebody that it is all we need to state faithfulness, OR is it just a component of faithfulness assessment.

Challenging of a scientific statement can be made two ways:
- Find a case where it is wrong
- Find an additional, unknown factor that changes the understanding

The latter means that:
a. You have to make the hypothesis that such a factor exists.
b. You have to make research in order to prove the hypothesis valid (no search, no finding, statement stands as is)
c. Negative result doesn't mean the hypothesis is wrong. It just proves the research did not confirm it (but another protocol could)

This is the name of the game in research.

Long story short, my hypothesis is that there are additional (dynamic) factors for faithfulness. For as long as nobody proves it true, I'm considered wrong. But for as long as nobody proves it wrong, I can legitimately make it.

Well, enough philosophy for today. Have a great day all of you

 

[hardisj]

A few keen eyes noticed I was testing the Kef R3 and caught my measurements, too, showed a dip ~1kHz. This matches everyone's response except Amir's. A couple PMs from two different people noting the shadow flare was sticking out a bit in my photos. Now, since I wasn't going "full bore" in to the Kef measurement just yet I hadn't paid attention to it. I received a pair. I knew I had already pushed one of the speaker's shadow flare in. But I hadn't checked the one on the stand yet because... well, I hadn't gotten there yet. I went to the garage to check and sure enough, the flare was sitting out about 1-2mm. Upon some investigation this is what I found:

Read the legend.

Bottom line is:
If the shadow flare isn't pushed all the way in there is a dip. One would think that flush would be the way to go. But nope. And when you look more closely it makes sense why. The shadow flare all the way in (leaving a small lip between the baffle and the mounting hole) makes the flare flush with the surround edge. Any position out from that - like flush with the baffle - leaves a small gap and causes the response to dip ~1kHz. The further out the shadow flare, the more drastic the dip. You want the shadow flare to meet the drive unit; not the baffle (well, both would be ideal but in a case of either/or, this is the route you go). And the data backs this up.


I hope this helps answer some of the questions asked here.

 

[Rick Sykora]

A few keen eyes noticed I was testing the Kef R3 and caught my measurements, too, showed a dip ~1kHz. This matches everyone's response except Amir's. A couple PMs from two different people noting the shadow flair was sticking out a bit in my photos. Now, since I wasn't going "full bore" in to the Kef measurement just yet I hadn't paid attention to it. I received a pair. I knew I had already pushed one of the speaker's shadow flair in. But I hadn't checked the one on the stand yet because... well, I hadn't gotten there yet. I went to the garage to check and sure enough, the flair was sitting out about 1-2mm. Upon some investigation this is what I found:

View attachment 112562
Read the legend.

Bottom line is:
If the shadow flair isn't pushed all the way in there is a dip. One would think that flush would be the way to go. But nope. And when you look more closely it makes sense why. The shadow flair all the way in (leaving a small lip between the baffle and the mounting hole) makes the flair flush with the surround edge. Any position out from that - like flush with the baffle - leaves a small gap and causes the response to dip ~1kHz. The further out the shadow flair, the more drastic the dip. You want the shadow flair to meet the drive unit; not the baffle (well, both would be ideal but in a case of either/or, this is the route you go). And the data backs this up.


I hope this helps answer some of the questions asked here.

Hi Erin. You might have flair, but think the KEF has a flare...

 

[hardisj]

Uh-oh.... we've got a spelling aficionado on our hands....




(fixed)

 

[napilopez]

A few keen eyes noticed I was testing the Kef R3 and caught my measurements, too, showed a dip ~1kHz. This matches everyone's response except Amir's. A couple PMs from two different people noting the shadow flair was sticking out a bit in my photos. Now, since I wasn't going "full bore" in to the Kef measurement just yet I hadn't paid attention to it. I received a pair. I knew I had already pushed one of the speaker's shadow flair in. But I hadn't checked the one on the stand yet because... well, I hadn't gotten there yet. I went to the garage to check and sure enough, the flair was sitting out about 1-2mm. Upon some investigation this is what I found:

View attachment 112562
Read the legend.

Bottom line is:
If the shadow flair isn't pushed all the way in there is a dip. One would think that flush would be the way to go. But nope. And when you look more closely it makes sense why. The shadow flair all the way in (leaving a small lip between the baffle and the mounting hole) makes the flair flush with the surround edge. Any position out from that - like flush with the baffle - leaves a small gap and causes the response to dip ~1kHz. The further out the shadow flair, the more drastic the dip. You want the shadow flair to meet the drive unit; not the baffle (well, both would be ideal but in a case of either/or, this is the route you go). And the data backs this up.


I hope this helps answer some of the questions asked here.

Thanks for investigating this Erin and confirming that I wasn't totally crazy!

I remember thinking at one point that it perhaps shouldn't be totally flush, as it struck me as odd that it seemed more people received the R-series speakers with a shadow flare that wasn't totally flush than those who did. But I didn't look into it as much as I should've when I had the speaker.

This was why Amir's spin was a bit surprising, because these were the only measurements I'd seen that didn't show the dip. It seemed Amir got the most perfect sample of all, as even KEF's measurements showed a tiny dip here.

Nonetheless, it's still pretty bonkers that such a small difference (we're talking a range of just 1mm!) could make such a noticeable and likely audible difference. I guess that goes to show why it's hard to make a really good waveguide too.

 

[hardisj]

Nonetheless, it's still pretty bonkers that such a small difference (we're talking a range of just 1mm!) could make such a noticeable and likely audible difference.

I agree. I didn't expect the response to be that dependent. But I was thinking in terms of the front of the baffle rather than in terms of aligning the flare to the speaker cone edge (and continuing that intended transition) because I figured the flare was thick enough to sit flush at both places; not just one. It appears they should have made this about 1mm thicker?

 

[theyellowspecial]

Is the flare the inner, middle or outermost ring? Is it meant to be adjustable?

 

[napilopez]

Is the flare the inner, middle or outermost ring? Is it meant to be adjustable?

The flare is the outermost ring, it extends the waveguide effect of the Uni-Q unit to prevent the driver from 'seeing' the baffle edges.

I agree. I didn't expect the response to be that dependent. But I was thinking in terms of the front of the baffle rather than in terms of aligning the flare to the speaker cone edge (and continuing that intended transition) because I figured the flare was thick enough to sit flush at both places; not just one. It appears they should have made this about 1mm thicker?

Could be! Just a small shame that there is such variation in this part. Seems slightly tighter tolerances could make this speaker even better than it already is for most people. Especially as ive seen so many people say they've pushed the flare to be flush with the baffle, when your measurements show that shouldn't be done!

I really liked the R3 but thought it sounded just a tad dull/laid back and believed that to be because of the 1kHz dip. Wonder if it would've been even better if I had the flare at just the right point.

 

[hardisj]

I really liked the R3 but thought it sounded just a tad dull/laid back and believed that to be because of the 1kHz dip. Wonder if it would've been even better if I had the flare at just the right point.

Dude! Me, too. I thought it might have been the room it was in because it is pretty dead. I wanted to see how this changed in the living room (less dead) but the same thing. I'm thinking this 1kHz dip was the culprit because I put in my notes "1-2kHz sounds thin" (relative to the nearby frequencies). I think that is what I was hearing.

 

[napilopez]

Dude! Me, too. I thought it might have been the room it was in because it is pretty dead. I wanted to see how this changed in the living room (less dead) but the same thing. I'm thinking this 1kHz dip was the culprit because I put in my notes "1-2kHz sounds thin" (relative to the nearby frequencies). I think that is what I was hearing.

I think it stands out because the rest of the speaker is so close to neutral!

 

[tuga]

Dude! Me, too. I thought it might have been the room it was in because it is pretty dead. I wanted to see how this changed in the living room (less dead) but the same thing. I'm thinking this 1kHz dip was the culprit because I put in my notes "1-2kHz sounds thin" (relative to the nearby frequencies). I think that is what I was hearing.

Or perhaps the narrower than usual directivity?

 

[hardisj]

Or perhaps the narrower than usual directivity?

I wondered that, too. It’s hard to know with absolute certainty. But that was my concern. Though, the DD8c might counter that. (I would have to check the radiation patterns of each to get a better idea of this)