• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

KEF LS50 Meta Review (Bookshelf Speaker) Video

Thanks to Amir and the members who very kindly took the time to respond to this post. My current understanding is

  • Preference scores are an objective way to rank speaker performance
  • There are two preference scores - with and without a subwoofer so you can pick the score that is appropriate to the configuration you wish to implement
  • Preference scores are computed using near field measurements. Given that most 2 channel hi-fi setups will be for midfield listening (defined as listening distances greater than 2 metres) you also need to factor in desired SPL at the listening position.
  • The above means that monitors with a woofer size smaller than 8 inches may be inappropriate for mid field listening irrespective of how good the preference score is
  • Powered monitors are better than passive monitors particularly if you wish to integrate a subwoofer


I am not sure that everyone will agree with the above but I thought it would be useful to set out this framework.
 
  • Preference scores are an objective way to rank speaker performance


Preference scores are based of a mathematical model / formula that has been developed by Olive et al. (you can read more about it here).

  • Preference scores are computed using near field measurements.


Quite the contrary, the Perference sore is almost 50% based on the predict in-room response, that's another mathematical model, it's quite complicated but basically when it was modeled it was off several averaged measurements with each measurement being 3 meters away from any other measurement. it's a room-wide predicition, not really near field or far field, but definitely not near field in anyway.

  • Powered monitors are better than passive monitors particularly if you wish to integrate a subwoofer

I'm not aware of any reason to assume so.
 
Thanks for the very interesting response. This contradicts some of the other members views.
So, in your opinion woofer size is immaterial?
Regarding the comment that powered monitors are better than passive ones. This was based on the ease of integrating a powered monitor with a subwoofer since the powered monitor will have a low pass cut off switch.
 
If you look at KEF's white papers, you will notice many, if not most, of their speakers have a similar downward slope in the bass frequencies. It appears they assume room gain will be present in most installations, which is probably a very good assumption.

They say exactly that in the original LS50 white paper:

It was found that a reflex-loaded enclosure with a somewhat over-damped alignment, tuned to 55Hz, allowed music to be reproduced at satisfying levels with some low bass without losing the impression of clarity. The gentle roll-off from below 100Hz also gives good flexibility to position the speakers in smaller rooms where the boundaries will provide some bass lift.
 
I had been looking at Genelec powered, and the passives Polk Audio's L100, & Golden Ear BRX. I wish there was a store where you can do a side by side. The KEF looks great.

me too, polk, genelecs, kef, revel. polk has R200 with measurements here..
 
Preference scores are computed using near field measurements. Given that most 2 channel hi-fi setups will be for midfield listening (defined as listening distances greater than 2 metres) you also need to factor in desired SPL at the listening position.
As explained, the preference score completely assumes far-field listening. Indeed we don't have one for near-field listing.

If you got that impression from "Near Field Scanner" that I use, I can see that but again, the reality is the opposite. The scanner uses near-field measurements to compute far-field results. Unless noted otherwise, everything I post is far-field.

For completeness, the standard stipulates "far field" to be 2 meters, reported at 1 meter. I have looked a number of speaker measurements and difference past 2 meters is negligible so it is as good as far fied. FYI I can generate results for any distance with my system. You may have noticed "10 meter" designation on directivity plots and such.
 
So, in your opinion woofer size is immaterial?

it's a factor, just not absolutely decisive.

This was based on the ease of integrating a powered monitor with a subwoofer since the powered monitor will have a low pass cut off switch.

A lot more is involved in integrating a subwoofer. Nevertheless, mid-end subwoofers usually offer a high-pass filter for speakers so that's not an issue at all.
 
Based on what Amir says, my conclusions now are
  • Use the preference scores to rank speakers.
  • These scores are valid for midfield listening situations which comprise the vast majority of home hi fi applications
  • If you intend to use subs, powered monitors make sub integration much simpler
  • And, if you intend to use subs, be guided by the preference scores with an ideal sub

Very clear indeed

Thanks for the clarification Amir.
 
Wondering out loud how Amir would compare to the less expensive Revel M16's
 
Based on what Amir says, my conclusions now are
  • Use the preference scores to rank speakers.
  • These scores are valid for midfield listening situations which comprise the vast majority of home hi fi applications
  • If you intend to use subs, powered monitors make sub integration much simpler
  • And, if you intend to use subs, be guided by the preference scores with an ideal sub

Very clear indeed

Thanks for the clarification Amir.

The first is not the correct conclusion, strictly speaking. Note that the preference score is held offsite and is not an official ASR score. Indeed, from what I can tell, the preference score project was abandoned by its creator, Sean Olive (someone please correct me if I am wrong), though it may have been tweaked a bit to adjust for some limitations by those that keep the databases that use them.

If you look at the preference score site, there is a section devoted to the formula and its limitations and affordances. Included are the following:

"This formula is based on far-field listening" -- This has been covered.

"Aspects such as sensitivity, distortion, difficulty to drive, max SPL, etc. are not factored in." -- This is very important to note and factors into the consideration of these speakers. All high ranking or similarly ranking speakers are not equal, in other words. Caveat emptor.

"This formula has a predictability of 0.86" -- I guess this might be correct within its limited context, maybe, but note that Amir has liked speakers with low preference scores and not liked speakers with high preference scores.

The preference score is, however, convenient if properly understood and therefore has value, but do not think that the ranking as you see it is indicative of a true official "ranking" by the audio community. Each speaker has its limitations and you are best off understanding as many of the measurements as you can in order to make the best decision for your specific use case.

TLDR: it's fine to use the preference score to get a ballpark idea of the quality of the speaker, but understanding the measurements and reviews are much more important.
 
Hi,
Many thanks for the response. While I completely understand the limitations of the preference scores is there anything better than the cloud cuckoo land of subjective reviews?
Specifically you state the limitations
"Aspects such as sensitivity, distortion, difficulty to drive, max SPL, etc"
With regard to the above,
Sensitivity – not an issue with powered monitors
Distortion – I assume this is covered in the near field measurements
Difficulty to drive – again, not a problem with powered monitors
Max SPL - Powered monitors typically have max SPLs of above 100 db. I think most people listen at 75 to 85 db so I do not see this as being an issue

Please note I am not trying to be argumentative here. I'm trying to get the "ideal speaker"defined. I think Amir has achieved this with electronics and I see speakers as being the final frontier

Just trying to help the ASR community which is an absolutely brilliant one
 
Max SPL - Powered monitors typically have max SPLs of above 100 db. I think most people listen at 75 to 85 db so I do not see this as being an issue

Max SPL tends to be the key limitation of powered monitors under $1000. Most are studio monitors made for near field listening at ~85db. Far field requires much higher SPLs. Read the specs carefully for the Max SPLs - its usually at 1 meter and well after compression has kicked in.
 
Hi,
Many thanks for the response. While I completely understand the limitations of the preference scores is there anything better than the cloud cuckoo land of subjective reviews?
Specifically you state the limitations
"Aspects such as sensitivity, distortion, difficulty to drive, max SPL, etc"
With regard to the above,
Sensitivity – not an issue with powered monitors
Distortion – I assume this is covered in the near field measurements
Difficulty to drive – again, not a problem with powered monitors
Max SPL - Powered monitors typically have max SPLs of above 100 db. I think most people listen at 75 to 85 db so I do not see this as being an issue

Please note I am not trying to be argumentative here. I'm trying to get the "ideal speaker"defined. I think Amir has achieved this with electronics and I see speakers as being the final frontier

Just trying to help the ASR community which is an absolutely brilliant one

100 db over 100 Hz.
 
@amirm , I had an idea for your back to school Youtube series. It'd be interesting to see a video explaining fundamentals, harmonics, evens, odds, first, second, third order etc... tied into how they relate and what they mean to distortion measurements, which are most important, etc... This terminology is very common in reviews, but I'd hazard a guess a lot of people (like myself) probably don't really know what it is or what it means in practice.
 
Excellent. The situation is becoming clearer and clearer. The issue boils down to SPL at the listening position. Fortunately there is a formula that calculates this. With every doubling of the listening distance there will be a 6 db drop in SPL.
Most powered monitors rate max SPL at 1 metre. Doubling this distance to 2 m will lead to a 6 db drop and doubling it again to get to a listening distance of 4 metres will result in a further 6 db drop.
This means, to cover listening distances between 1 m and 4 m (this should cover the vast majority of foreground listening) will lead to a 12 db drop in max SPL.
So, as long as the rated SPL of your chosen powered monitor less 12 sb gives you the SPL at which you normally listen, powered monitors with or without a sub would be an excellent solution
 
"This formula has a predictability of 0.86" -- I guess this might be correct within its limited context, maybe, but note that Amir has liked speakers with low preference scores and not liked speakers with high preference scores.

I think this may depend on what you mean by "predictability", at least from what I understand. I know @preload has pointed out that .86 correlation means that it predicts 74%(.86^2) of the differences in scores. I may be saying that wrong, though.

That's also the figure for the full range test. The smaller test where bass was equalized actually showed much better correlation between measurements and predicted preference(.99 correlation).

With the database we're building now, we could probably start improving these figures via blind tests of measured speakers.
 
I think this may depend on what you mean by "predictability", at least from what I understand. I know @preload has pointed out that .86 correlation means that it predicts 74%(.86^2) of the differences in scores. I may be saying that wrong, though.

That works. Also, the correlation factors in bass response as well.

That's also the figure for the full range test. The smaller test where bass was equalized actually showed much better correlation between measurements and predicted preference(.99 correlation).

No idea what this is referring to.
 
Hi Richard,
Thank you for the response. Very helpful to be advised of the predictability percentages. Ultimately, as I have said before, while this is not a perfect measure it is a million times better than subjective reviews.
Double-blind listening tests would take the scores to an entirely new level of accuracy and I hope you succeed in this endeavour. I am based in south-east England and I'm sure there are many ASR members who live nearby.
Please let me know if I can help in any way
Kind regards
 
Amir,

A question about some of your opening comments on the Spinorama. You state that the on-axis sound is usually the best from most speakers and that is how they should be listened to. I don't find that to be the case, and most of your measurements don't show that, either. In most cases, the listening window is flatter and smoother than the on-axis (0°) sound, indicating a 10-30° angle being the best. In my own experience, pointing the speakers I've owned directly at me has been harsher than I care for. Can you comment further on this?
 
Last edited:
I don't find that to be the case, and most of your measurements don't show that, either. In most cases, the listening window is flatter and smoother than the on-axis (0°) sound, indicating a 10-30° angle being the best. In my own experience, pointing the speakers I've owned directly at me has been harsher than I care for.
I think exactly the same. Some speaker brands (Dali, for example) intentionally design their speaker to sound their best off-axis: say, up to ≈-30° horizontal.

As far as I'm concerned, I used most of my speakers with zero toe-in, such as my actual A500s: they show a very even listening window but FR is clearly not that flat if you look strictly at 0°. In fact, I found most, if any, speakers to sound harsh rightly pointed at me.

There're sure some exceptions, like the dead flat Gens 8030 I listened at @daftcombo 's with no harshness to speak of even right on axis.
You state that the on-axis sound is usually the best from most speakers and that is how they should be listened to
I usually look at PIR to be one of the most significant graph in speakers reviews.
 
Back
Top Bottom