Subjective Comparison and Conclusion
TLDR
There is no clear winner here. Both speakers are excellent and do some things better than the other. I like them both very much, and after 13 months with the BMRs and 1 month with the R3s, I have not made a decision as to which speaker stays on the stands.
BMR has wider dispersion and casts and immensely wide soundstage
BMR has ~6Hz lower bass extension, going as low as 27Hz in my room at meaningful SPL
BMR has sparkling highs that sound less bright overall
R3 works better in live rooms, but casts a narrower soundstage
R3 soundstage is deeper
R3 is less sensitive to tweeter height placement, but sounds seem to emanate directly from the tweeter in some cases
R3 is brighter above 6KHz
Horizontal Directivity
I do not know that we have good spins of the BMR to show its directivity characteristics, but horizontal dispersion sounds very wide. Both the RAAL ribbon and BMR midrange react with the side walls to create a wide soundstage. I cannot treat the side walls in this room, and there have been times I wished I could.
We do have good spins of the R3 and know horizontal directivity is narrow-ish, but very well-controlled. The audible soundstage matches the measurement, and this can be a good thing in live rooms such as mine, as it never sounds confused. That is not to say the soundstage is narrow, however; it is plenty wide.
Both speakers are excellent at creating and maintaining a phantom center image.
Neither speaker shows a clear disadvantage in frequency response when measuring common toe-in angles in my room, as the delta is < 1dB between 0 and 2" toe-in, even at high frequencies.
Because directivity is room-dependent, we have a tie on this point, however the nod may go to the R3s in my room.
Vertical Directivity
The BMRs are interesting in this regard. I found you do not want to sit more than 10 degrees above or below the RAAL tweeter. Above sounds dry, and below creates an odd image, where the universal directivity of the BMR midrange competes with the diminishing output of the tweeter to create an unpleasant sound field in the upper mids. I solved this quite easily by purchasing the correct height stands for the speakers. Having said that, once the dangly things on my melon were in the correct orientation to the tweeter, the BMRs cast a satisfyingly tall soundstage in my room.
The R3s are mostly agnostic to vertical placement due to the coaxial driver. However, when the R3s were placed on the BMR's 24" stands, which placed the tweeters well below my ears, some material sounded low and collapsed in vertical soundstage. Once such recording is 10,000 Maniacs MTV Unplugged. I have no explanation for why this happened with some recordings and not others. After swapping the stands for 32" stands, which align the tweeters almost exactly with my ears as I sit in an office chair, the problem disappeared, and the affected recordings sounded normal.
Both speakers have their idiosyncrasies, and both are easily fixable. It's a tie.
Depth
This is one of those nebulous attributes around which I cannot wrap my mind. Some recordings sound deeper than others, and some speakers convey that depth better than others. The only possible explanation I can think of is the ability of a speaker to layer instruments more accurately(?). A good recording for testing this is The Corrs Unplugged.
The only thing I can really say is the BMRs are fine in this area, and the R3s are outstanding. This is something you have to hear for yourself, and it is nearly impossible to describe without resorting to word salad comprised of audiophile nonsense words.
R3s score a point here.
Frequency Response
OVERALL
The posted graphs demonstrate the R3s have a smoother in-room response throughout the > Schroeder range in my room. I have to listen very carefully to hear any differences between 1000Hz and 6000Hz, meaning the wiggles in the BMR response are likely inaudible to most people. Overall, the BMRs sound warmer. Both speakers sound mostly balanced, save for the delta above 6KHz.
BASS
One attribute people love about the BMR is its ability to play below 30Hz with authority. This makes subs unnecessary for many people.
The R3 reaches an admirable F3 in-room as well, however it is ~6Hz higher than the BMR. People often comment on low bass SPL when examining spins of KEF speakers, but I have not found them to behave any differently with typical placement in a typical room. The measurements in this thread back that up.
Beyond low frequency extension, both speakers produce usable SPL in the bass region.
Point to the BMRs.
MIDRANGE
Here, the speakers sound more similar than different. Both play clean and clear at the SPL I use in that room. Both exhibit smooth FR in the midrange region. Both sound great. There isn't much more to say.
Another tie.
TREBLE
For context, my own hearing tests indicate I can hear 23 to 14,800Hz through speakers in this room.
We have seen the BMR has some wiggles in the treble region as it rolls off in-room. The dip at 8KHz, for example, is audible with the right material playing, assuming I bother to concentrate and listen for it. Otherwise, treble generally follows the roll off of the Harman curve. The RAAL ribbons are special units with accurate tonality and some added sparkle (I guess). These are very nice tweeters.
In contrast, the R3's treble is smoother in FR, but is also elevated above 6KHz, and this is audible, and obviously so, with most of the tracks I have auditioned thus far that have content up there. This skews the R3's tonality toward bright. I have mixed feelings about this. I appreciate the help above 15KHz, but the region between 6KHz and 15KHz calls attention to itself, and I can see how some people would be torn between describing it as "detailed" and "fatiguing". To compensate for this extra HF energy, I made two Dirac presets: limited to 1KHz correction and full range. The first preset is the default. I hit the second one when an album sounds overly bright. Pointing the speakers straight ahead helps by between 0.5 and 1dB vs. 1.5" toe-in, therefore it is not a perfect solution. As for tonality, all I can say is treble sounds smooth without any noticeable flaws.
BMRs score another point.
Dynamics
This comparison is purely subjective on my part. The only way I can qualify dynamics is to report on whether one speaker or the other is obviously deficient, and neither speaker is. They are both fine according to what I can hear in compression and distortion. Loud passages sound loud.
All I can do is call it another tie.
Sensitivity
BMR sensitivity is listed at 86dB (2.83V/1M) with nominal impedance of 4 Ohms.
R3 sensitivity is listed at 87dB (2.83V/1M) with nominal impedance of 8 Ohms.
I found both speakers to want similar power from my amplifier. I cannot comment on lower power amps with these speakers other than to say my former Cambridge CXA60 drove the BMRs with no problems. My current amp was measured by Amir and pumps out enormous amounts of power (~450Wpc @ 4 Ohms reserve) for my application, so ability to drive any given speaker is not a concern for me.
The two speakers were within a decibel or two of each other with the same volume setting as measured in my room. Hitting volume up once or twice brought the R3 up to the same SPL as the BMR.
The BMR technically wins this one, but it is functionally a tie in my application.
Conclusion
As mentioned above, there is no clear winner. Each speaker has its pros and cons that may make it more desirable in a given environment.
Both feature high quality drivers, well-made inert cabinets, and effective crossovers.
For my part, I will keep the R3s on the stands for a while to see if they make the grade over the long term. I like the narrower directivity in this room. I also like the depth. I am not a fan of the hot treble so far. Time will tell.
The BMRs will be placed into foster care with a friend to see what he thinks of them, as he is considering BMR towers or something else from Salk Sound. Once I get them back, I may strip and refinish the cabinets to something that fits better with the decor of our home.
[Is what I am calling depth, and Erin called 'layering' in his review the reason I at least believed I missed these speakers? If so, is that something that shows up in measurements? Or is this a subjective angle that perhaps should be considered in speaker selection?]