KEF R3 Speaker Review

[echopraxia]

I believe some of what appears to be a disconnect between the excellent measurements of the R3 and our host's preference for the M16 which has very good measurements but noticeably below the R3 is we are looking at the preference on one person. The research has shown if a group of people blind tested these two speakers more would prefer the R3 to the M16. Of course, it would not be every single listener in the test.

I would love to see such a large scale blind test experiment performed! Perhaps not necessarily with the KEF R3 vs Revel M16, but any such similar pair of two speakers where it seems the predicted preference (from measurements) doesn't seem to align with reality. For example, what if we find that the spins show that a KEF R11 should be preferred over a Revel Salon2? If such a prediction is made from the preference score, I think it would be valuable to actually take such an example and test it thoroughly in a blind test.

I'm a little less certain than some here that the preference scores we're seeing will perfectly predict the outcome of such a large-scale blind test for every pair of speakers, perhaps because I work in an industry where we fit statistical models / train machine learning models all the time -- and I understand and have seen first-hand the plethora of ways in which such models can fail to make reliable predictions outside of the set of data the model was fit to.

I could write extensively about statistical models vs objective science, but in short: It seems that many fellow objectivists here forget that while the raw measurement data is objective, preference scores are not objective truth, they are statistical models. No matter how good those models are, there is likely some room to improve them.

But again, regardless of how we interpret them regarding predicted speaker preference, the measurements themselves are objective and intrinsically valuable -- and I'm super excited to watch as this site builds an increasingly large collection of high quality data here!

 

[maty]

https://www.diyaudio.com/forums/mul...und-measurements-raw-drivers-post5863878.html

-> https://www.hifinews.com/content/kef-r3-loudspeaker-lab-report

KEF R3

Resized

Sensitivity (SPL/1m/2.83Vrms – Mean/IEC/Music): 86.1dB / 86.1dB / 86.1dB

Impedance modulus min/max (20Hz–20kHz): 3.1ohm @ 44Hz / 13.1ohm @ 2.4kHz

Impedance phase min/max (20Hz–20kHz): -42° @ 37Hz / 32° @ 1.3kHz

Pair matching/Response Error (200Hz–20kHz): ±0.7dB / ±2.0dB / ±2.1dB

LF/HF extension (–6dB ref 200Hz/10kHz): 40Hz / >40kHz / >40kHz

THD 100Hz/1kHz/10kHz (for 90dB SPL/1m): 0.5% / 0.2% / 0.1%

Dimensions (HWD) / Weight (Each): 421x200x312mm / 14kg

Price: £1300

 

[tecnogadget]

Not sure how you come to this conclusion, having experience of KEF coax speakers my experience more or less reflects @amirm ’s.

The thing is that the current UniQ used in the R3 is the 12th iteration/generation. So not sure how it can correlate to those previous experiences with old models. And in between some generations you get different tiers (budget, mid, reference, etc)

 

[maty]

Where is the dip about 1 kHz in Klippel graph?

 

[MerlinGS]

Another consideration is age. Dr. Toole no longer participates in speaker evaluations because he feels his decline in hearing acuity may make him a poor choice (I'm paraphrasing something he wrote at AVS). How old is Amir? I would assume it is a reasonable guess to suggest he has lost some of his hearing acuity. Could the loss be a factor reflecting his choices? Also, as others have suggested, Amir's many years of exposure to Harman's "house" sound could arguably predispose him to prefer their design preferences (Harman makes excellent speakers, Amir has been aware of their excellence for many yrs, and this could reflect his choice or it could lead to bias in his choice [there is a reason Harman does speaker comparisons blindly]). Lastly, as Ron Texas noted, the Harman tests do not suggest that all trained listeners will choose the same excellent speaker, but rather the majority of listeners are more likely to chose the "better" measuring speaker. To my knowledge neither the scientists at Harman nor Amir has ever suggested the preferential measurements would indicate some objective truth, but rather it would indicate a preference that had statistical significance (i.e. could not be used to predict a particular individual would prefer the speaker, but rather it could serve to predict which speaker the majority of trained listeners were likely to prefer in a blind test). Two trained listeners with excellent hearing and evaluating skill could consistently prefer to different types of speaker designs.

 

[laurelkurt]

Time ago I found the crossover of the previous model, R300. Quite disappointing (for the range price) the components quality. Presumably R3 will be similar.

KEF R300 crossover, pictures by Zvu

https://www.diyaudio.com/forums/mul...und-measurements-raw-drivers-post5520105.html

View attachment 54154

KEF R300, inside

https://www.diyaudio.com/forums/mul...und-measurements-raw-drivers-post5590785.html

Can expensive x-over components actually make a measurable difference in the sound as long as nothing is being overloaded?

 

[tecnogadget]

Would this be good enough for the price point ? R3 are £1,299.00 (like US$1550) in UK...Usa price is higher because of imports

 

[aarons915]

I just don't have the time. Someone just got anxious that I had not tested their gear and I had to ship it back without testing. The purpose of these reviews is to post the objective data. Other side projects will have to be performed by someone else as I work through the mountain of hardware waiting to be tested.

100% agree with this logic, it would be a shame to have people start getting impatient and wanting their gear back before testing. My suggestion until the backlog is measured is to simply measure the speakers and post the basic Spin info like in this review, attach the data for those interested and don't even do listening tests unless it's something you're really interested in comparing.

 

[amirm]

100% agree with this logic, it would be a shame to have people start getting impatient and wanting their gear back before testing. My suggestion until the backlog is measured is to simply measure the speakers and post the basic Spin info like in this review, attach the data for those interested and don't even do listening tests unless it's something you're really interested in comparing.

Thanks. Listening tests are very important to me personally as it allows me direct comparison between what I have measured, and what I perceive. It is a great form of training so unless time pressure is high, I will continue them.

 

[ROOSKIE]

Can expensive x-over components actually make a measurable difference in the sound as long as nothing is being overloaded?

Maybe, so far only subjective impressions have thumbed up expensive crossover parts. Measurements do not show changes as of yet.

 

[infinitesymphony]

If I did most of my listening on a pair of Revel Ultima Salon2s I would probably vote for whichever bookshelf speaker had more bass represented.

Amir's speaker review is a freebie on top of comprehensive (and expensively-obtained) objective results. I guarantee that he understands what would be required to properly conduct blind ABX speaker testing and it goes beyond the current scope of this project.

If cost is a deciding factor, that says a lot about the Revel M16. Could there be something lurking in the $1-2K/pair range with similar or better results than the KEF R3? My vote: Neumann KH 120.

 

[amirm]

Where is the dip about 1 kHz in Klippel graph?

Crossover frequency is 2.9 kHz or quite a bit higher than 1 kHz. Why should a dip be there?

There is not one in KEF measurements either.

 

[napilopez]

I believe some of what appears to be a disconnect between the excellent measurements of the R3 and our host's preference for the M16 which has very good measurements but noticeably below the R3 is we are looking at the preference on one person. The research has shown if a group of people blind tested these two speakers more would prefer the R3 to the M16. Of course, it would not be every single listener in the test.

Not sure how you come to this conclusion, having experience of KEF coax speakers my experience more or less reflects @amirm ’s.

I expect that we @ASR as we gain more experience will become the research setting the standard, we are the future in audio I strongly believe.

I think @Ron Texas right on the mark. I think it's pretty wild to think that one user's sighted, largely uncontrolled preference will correlate exactly with the preference score derived from hundreds of listeners, tested in one room, with an accuracy of 86 percent, under very specific conditions.

We can extrapolate things from that test into the real world, but it doesn't mean limited individual experience will correlate precisely back with it. Especially considering that though the Revel M16 scores worse, it's still in the upper echelon of speakers. Dr Toole calls for flattish on axis and smooth directivity. Both of these speakers achieve that.

I'd be more concerned if Amir preferred, say, the JBL Control 1. Personally, I've heard plenty of KEF Coaxes and have liked them all, with the R3 being the best of the bunch (I've not heard the higher-end models).

 

[thewas]

If a speaker gets to end up with an In-Room response curve that represents a broad rise after a 1k-2k depression can be perceived as a little bit anemic and bright (R3). And if it gets some rise up in the presence region like 1k to 3k or has some step hard knee in the sub-bass/bass region you get the sense of fuller body, more pleasing tone, less fatigue (M16, R2C). I dunno, but this is something that could correlate.

That's also what I also think and wrote here.

 

[Senior NEET Engineer]

I think @Ron Texas right on the mark. I think it's pretty wild to think that one user's sighted, largely uncontrolled preference will correlate exactly with the preference score derived from hundreds of listeners, tested in one room, with an accuracy of 86 percent, under very specific conditions.

It is 99% accurate for bookshelf speakers with restricted low frequency output.

 

[amirm]

While waiting for a new speaker finishing number crunching, I ran this comparison between the KEF R3 and Revel M16:

These are tweeter axis, "in-room" measurements with no signal processing (i.e. have room effects in them). The "room" is my garage where the Klippel lives. I have compensated for level difference by making both the same at 1 kHz.

It seems that the KEF R3 in-room is NOT following the harman target curve of sloping down. If anything, it peaks back up.

The Revel is also flatter than prediction but it is closer to its predicted target than the R3 is.

I plan to keep the R3 and perform some more tests along these lines......

 

[aarons915]

While waiting for a new speaker finishing number crunching, I ran this comparison between the KEF R3 and Revel M16:

View attachment 54168

These are tweeter axis, "in-room" measurements with no signal processing (i.e. have room effects in them). The "room" is my garage where the Klippel lives. I have compensated for level difference by making both the same at 1 kHz.

It seems that the KEF R3 in-room is NOT following the harman target curve of sloping down. If anything, it peaks back up.

The Revel is also flatter than prediction but it is closer to its predicted target than the R3 is.

I plan to keep the R3 and perform some more tests along these lines......

You didn't note any harshness in your listening tests but when I brought home the R3 to audition in my room they fatigued me within an hour, otherwise they sounded good. I think it's most likely the rise you speak of, I don't think it's a good idea for the early reflections to decline and then rise back up the way they do either. These measurements make me curious about using a bit of EQ to see if I could tame them a bit, it doesn't seem like it would take much to make these amazing speakers.

 

[McFly]

True, which I just did it for fun, EQing the listening window of the R3 (only 5 PEQs needed) and with this EQ also the predicted in room response becomes very nice, showing a bit of constant directivity behaviour above approximately 1kHz:

View attachment 54116

If someone wants the PEQ coefficients just send me a PM.

Mate post those up those look nice

 

[infinitesymphony]

It seems that the KEF R3 in-room is NOT following the harman target curve of sloping down. If anything, it peaks back up.

So in your room there's a 2-4 dB difference all the way through the sensitivity range (2-5 KHz) and beyond with the R3s having a boost and the M16s having a cut. With the M16 bass bump it's easy to imagine that the R3s would sound fatiguing by comparison.

 

[amirm]

So in your room there's a 2-4 dB difference all the way through the sensitivity range (2-5 KHz) and beyond with the R3s having a boost and the M16s having a cut. With the M16 bass bump it's easy to imagine that the R3s would sound fatiguing by comparison.

This is not my listening room. It is the measurement room/lab. Finding a quiet time to measure my room is challenging.