Best measuring speaker?

[DanielT]

It depends on what criteria you have. Let's say the highest SPL with a minimum of 1% (perhaps a maximum of 5-10% below 100 Hz) distortion and FR and frequency range that is (you can think about what boundaries it should be within). My guess is that there is some PA speaker which then becomes "Best measuring speaker" compared to the suggested HiFi speakers in the thread.
For most people with home hi-fi listening that maximum SPL with minimum distortion would not be a criterion to go by. It would probably be really big PA speakers that you could play the crap out of. After that, you can visit the hearing doctor. That is, unless the neighbors (if you live in an apartment building) beat one to death before this happens. Heh heh.

On the contrary, no one would use a couple of small, very good measuring, HiFi speakers in a PA context.

 

[Kvalsvoll]

That's why I mentioned it. I see in many, many discussions people saying they like really wide dispersion because it "sounds the same all over the room," and is good for multiple listeners for that reason. And I think they're correct in many cases when tonality is the only thing they judge. But I do think some who say that haven't paid enough attention to the spatial aspect of the recording and how the soundstage almost completely collapses into the nearest speaker in such a setup.

Exactly. I also find speakers with good pattern control fill the room in a much better way, most important for the main lsitening position, but it also works much better for other locations.

 

[hvbias]

Kef Blade Metas are there, but you have to change the URL to show all data sources (not just those from Klippel NFS).

Use this link instead:

Ranking table of speaker's measurements

www.spinorama.org

Can someone link to KEF's full complement of measurements for the KEF Blade Meta 1? I'm unable to find it on their website. Thanks.

 

[sifi36]

Can someone link to KEF's full complement of measurements for the KEF Blade Meta 1? I'm unable to find it on their website. Thanks.

It’s in the white paper.

Link

 

[BossBunos]

I'd just like to add to the beginning of this "if you have twenty years to spare". There is much talk of how important a flat frequency response is or low distortion and so on but, as you say, a speaker is not one measurement or even many measurements, but the interaction of all those measurements and the individuals room (and amplifier, if passive). This is, presumably, very complicated, much more so than measurements might suggest.

I have asked a few times for some kind of hierarchical system as when you should prefer x (say frequency response) over y (low distortion) or how two or more measurements interact with each other, but have drawn mostly blanks, am I correct in thinking there are no real shortcuts here?

I find what Earl Geddes has to say about how audible certain distortion harmonics are very illuminating. I can't speak to its veracity, as I don't know enough, but I definitely would like to see more practical information presented on ASR, rather than 'here are the graphs, get on with it'.

I think the 'decipher for yourself' approach is liable to lead to misunderstandings for anyone not educated to a high degree in acoustics or similar - which is what, 5% or fewer forum members?

I think Amir mentioned a couple of times that the on-axis response is the most important. I believe that is because of how our brain processes sound. The direct sound weighs more than the reflected sound.
The second thing I look at is the directivity index (or estimated in-room response since they are related). If the reflected sound follows the same trend as the direct sound. You might argue this is more important than the direct sound if you will use EQ. EQ only changes the direct sound. So with a good directivity index you can fix the speaker. If the DI is not good the speaker can't really be EQ'ed. The Kef R3 is a great example. Without EQ it's a bit bright and lacks a bit of bass (from the top of my head) but it's DI is almost perfect. So a bit of EQ fixes the issues it has.
The other parameters are quite personal. Do you want bookshelfs with sub(s) or floorstanders? Personally I like floorstanders, play music quite loud and love deep bass. So the third thing I look at is bass extention and %distortion @96dB.

But like I said I think it will become pretty personal after the first two points.

 

[Travis]

I disagree. First, as its name tells us, it’s the best preferred speaker, not best measured. As you correctly pointed out the score is calculated using a limited set of measurements. There cannot be a best measuring speaker because a speaker design is always a compromise. You cannot have all available parameters the best.

You can disagree all you want, but you are going down a road of semantics and not useful analysis. There can be all sorts of best measuring speakers. The question is what are you measuring.

Sorry so late, but from my understanding of what the scores are all about, I have a question.

What precisely are the scores supposed to show, the number? The score is from Dr. Olive’s algorithm with the 87% correlation? Has he discussed what you can draw from two speakers within 1 full point of another? Or was that just headphones that he said too much weight is given to the score within a certain range of one another.

I ask because I have seen various responses about “scores” here. Some have been in the vein of they can get you to a point generally, but these scores shouldn’t be given the same weight like a specification (horsepower, torque for example, sensitivity, power handling). Other responses, not either of the two I quoted above, have been more in the vein of buying the highest scoring speaker, in a class of speakers, like floor standing, will get you the best Sounding in their room.

@Bjorn (who is quite knowledgeable in all of this from reading his posts here, and other forums for last ten years) and also others in this thread have pointed out they don’t tell you everything.

While measurements, score, most preferred, etc. are semantics, the thread’s original question is what is “best measuring speaker.” I understand that is obviously going to depend on the person’s weight he gives to any particular measurements.

So what precisely does the score of, let’s say, the 5 highest scoring speaker tell you about those five among each other? What DON’T they tell you? Is it that they will be the most highly preferred? That the speakers justify the price. What does it tell you about about No. 5 compared to No. 3?

What do the scores tell you if you don’t plan to run a sub? If you don’t want to run outboard DSP/eq.? If they are going to sit on a desk, or small listening room on stands, or large American living room?

Is there a minimum number for good sound? Or do we simply want something in top 100, 25, 10, 5, 3?

 

[Travis]

better
- a website that have a perfect methodology for ranking speakers

What’s the methodology and what does the rank tell you?

 

[pierre]

What’s the methodology and what does the rank tell you?

you can sort by various criteria: price, frequency flatness, spinorama, f3, f6, size, etc
the ranking gives you some broad information and as always how you interpret it is up to you.

 

[Travis]

you can sort by various criteria: price, frequency flatness, spinorama, f3, f6, size, etc
the ranking gives you some broad information and as always how you interpret it is up to you.

So they are meaningless? Completely up to subjective interpretation?

That can’t possibly be the case if tha score is based on the patented algorithm of Dr. Olive which has close to a 90% correlation.

I went to spinorama.org, the help page and am slowly going to the references, videos by Amir and others, but have come across an explanations of the scores yet, what they tell you with speakers with similar scores in relationship to one another, and what they don’t tell you.

 

[pierre]

So they are meaningless? Completely up to subjective interpretation?

That can’t possibly be the case if tha score is based on the patented algorithm of Dr. Olive which has close to a 90% correlation.

I went to spinorama.org, the help page and am slowly going to the references, videos by Amir and others, but have come across an explanations of the scores yet, what they tell you with speakers with similar scores in relationship to one another, and what they don’t tell you.

Hello,

the score is providing good information in the context of the study.
High score means:
- flat on axis, listening window and predicted-in-room-response (and as a consequence linear DI)
- low f3
- likely large horizontal dispersion

The score tells you nothing about distorsion, compression, or max SPL since they are not parameters.
The study was done a long time ago when divergence between good speakers was higher than now.
Precision (aka scores are statistically different) is around 0.6.

Here is my personal interpretation:
- Below 4 there is no reason to look at a speaker
- DI is more important than on axis flatness if you use a DSP. I prefer a linear DI speaker that I can equalize than the reverse. If I don't use a DSP I prefer flat on axis.
- Horizontal and vertical patterns are important (MTM vs Dipole etc)
- Max SPL or bass extension depends of what you need (big room big speakers)
- All the high score speakers are good and the differences between them decreases (Kii v.s. KEF v.s. Genelec v.s. Neumann etc)
- Large dispersion v.s. narrow depends a lot of the room and if you have multiple seat to cover. In a reverberant room, I prefer narrow and deal with the small sweet spot.

Note that similar score does not mean similar speakers: you can have a score of 5 for a large speaker with low f3 and a score of 5 for a nice baby coaxial. I have a page that shows speakers which are likely to sound similar (close spinorama), see here.

More technical argumentation:
- the model for the PIR works relatively well is a rectangular room but it is not clear to me if EQ should be on Listening Window or on the PIR. Near field I use the LW. Far field, TBD.
- the model itself is not good enough: some parameters needs to be bounded artificially or you get stupid answers when you optimise for the score
- optimising the score is mostly flattening the PIR which gives very bad results if the DI is not linear.
- this has been debated to death in a long thread here

End of the day, it is an interesting information about a speaker than kind of summarize the spinorama but of course not fully. That's also as far as I know the best and most scientific study around.

 

[quattro98]

Here's my personal, non expert take.

I started by understanding and identifying what mattered to me. This included things like bass extension/planned use of subwoofers, listening distance, listening levels, and other factors.

For me, I don't listen very loud (something like 70-75 dB) and at moderate distance of 10 feet. I wanted speakers as part of a multichannel system, so having a good center channel option was important to me. I also prefer floor standing speakers instead of stand mounts for more bass extension & sensitivity in a given footprint. I also prioritized even off axis response, both vertical and horizontal. Our system is in the living room and we have a sofa in front and one at 90 degrees (basic L shape). Especially when watching TV or streamed shows, we might get up and walk around or lie down on the sofa.

Given my preferences for multichannel options and good off-axis response, I wanted coaxial speakers. There are lots of options with high scores from KEF & Genelec. We wanted passive for ease of surround wiring. That basically narrowed it down to the R Meta & Reference Meta lines. From there, I no longer looked at the scores. I then looked at things like low distortion, low compression, size, build quality, appearance, and country of origin.

There are tradeoffs. Other speakers have greater horizontal directivity in the highs. Others might have narrower bass directivity (cardioid designs, etc.). If those are more important to you, then you can choose a different speaker with roughly the same scores and get what works for you.

 

[Tom C]

Hopefully, based on the excellent responses above, readers have a pretty good idea of the limitations of the score, and what people mean when they say that there are many factors to consider when choosing the best speaker for your own purposes, including important features that are not captured in the score. It is my understanding that the reason the score was developed was to assist designers in understanding what factors go into making a speaker that most listeners like. It was not intended to be a consumer shopping tool. So it should be no surprise that the score does not function perfectly as a shopping comparison tool, because it was never intended to be such. Some folks try to use it as a quick and easy way to find the ”best speaker,” but that’s really not what it is.

 

[Travis]

Hello,

the score is providing good information in the context of the study.
High score means:
- flat on axis, listening window and predicted-in-room-response (and as a consequence linear DI)
- low f3
- likely large horizontal dispersion

The score tells you nothing about distorsion, compression, or max SPL since they are not parameters.
The study was done a long time ago when divergence between good speakers was higher than now.
Precision (aka scores are statistically different) is around 0.6.

Here is my personal interpretation:
- Below 4 there is no reason to look at a speaker
- DI is more important than on axis flatness if you use a DSP. I prefer a linear DI speaker that I can equalize than the reverse. If I don't use a DSP I prefer flat on axis.
- Horizontal and vertical patterns are important (MTM vs Dipole etc)
- Max SPL or bass extension depends of what you need (big room big speakers)
- All the high score speakers are good and the differences between them decreases (Kii v.s. KEF v.s. Genelec v.s. Neumann etc)
- Large dispersion v.s. narrow depends a lot of the room and if you have multiple seat to cover. In a reverberant room, I prefer narrow and deal with the small sweet spot.

Note that similar score does not mean similar speakers: you can have a score of 5 for a large speaker with low f3 and a score of 5 for a nice baby coaxial. I have a page that shows speakers which are likely to sound similar (close spinorama), see here.

More technical argumentation:
- the model for the PIR works relatively well is a rectangular room but it is not clear to me if EQ should be on Listening Window or on the PIR. Near field I use the LW. Far field, TBD.
- the model itself is not good enough: some parameters needs to be bounded artificially or you get stupid answers when you optimise for the score
- optimising the score is mostly flattening the PIR which gives very bad results if the DI is not linear.
- this has been debated to death in a long thread here

End of the day, it is an interesting information about a speaker than kind of summarize the spinorama but of course not fully. That's also as far as I know the best and most scientific study around.

Thank you so much for that information!

 

[Travis]

It is my understanding that the reason the score was developed was to assist designers in understanding what factors go into making a speaker that most listeners like. It was not intended to be a consumer shopping tool. So it should be no surprise that the score does not function perfectly as a shopping comparison tool, because it was never intended to be such

That great information to know.