Master Preference Ratings for Loudspeakers

[MZKM]

Also, I wonder if cardioid response in the standing wave frequency range (ME Geithain) is considered in this compound score.

That will effect Sound Power and thus may effect SM_PIR. The formula cares about smoothness/deviation >100Hz, so a lower sound power due to cardioid cancellation may cause this if it enters that region.

 

[Krunok]

I think more speakers need to be measured before questioning why a $1000 speaker doesn’t top the charts.

This has nothng to do with price. As I said, i seriously doubt any speaker targeted toward home market, including Salno2, can match KH80 in terms of linearity.

I can’t alter the formula, but I could create a separate ranking; don’t know if it’s should be curved to add the score difference between the current best from a 10, or simply rank them numerically (just sort them by score and report their standings). both of these would of course be dynamic. I’m leaning the latter as it won’t give the impression of a pseudo-10 being the best out there, but instead the best measured so far.

I would wait until @amirm measures his Salon's.

 

[MZKM]

This has nothng to do with price. As I said, i seriously doubt any speaker targeted toward home market, including Salno2, can match KH80 in terms of linearity.



I would wait until @amirm measures his Salon's.

Or gets his hands on the D&D 8C.

Again, linearity on-axis is just one aspect.

 

[Krunok]

Again, linearity on-axis is just one aspect.

I am speaking about linearity of all 5 graphs.

 

[MZKM]

I am speaking about linearity of all 5 graphs.

And even their own graphs and the German magazine one shows that horizontal performance bunches up ~900Hz and vertical performance beams ~2kHz.
Now, how big of a knock-down in score should that give, I don’t know; but how big of a knock-down in score it does give is as shown (again, unless I calculated some things wrong).

 

[Krunok]

And even their own graphs and the German magazine one shows that horizontal performance bunches up ~900Hz and vertical performance beams ~2kHz.
Now, how big of a knock-down in score should that give, I don’t know; but how big of a knock-down in score it does give is as shown (again, unless I calculated some things wrong).

True. Let's wait for Salon2.

 

[Frank Dernie]

I'm aware of that. But if we can't find a single speaker that will breach the 8 mark than we can start questioning Olive's rating system to be too "strict".

I tend to take the opposite view.
Speakers are by a country mile the worst performing part of a hifi. If one of the small number of modestly priced units tested so far had a score of over 8 it would mean that the scale would be far too flattering IMHO, not stict enough in other words.
I wouldn't expect anything I have seen ever to deserve 9 out of 10 yet. There is a very long way to go before we see anything even close to perfection in speakers.
Lets see what a state of the art 3-way can do, and I still would not expect state of the 2020 art to deserve 90% personally

 

[Krunok]

I tend to take the opposite view.
Speakers are by a country mile the worst performing part of a hifi. If one of the small number of modestly priced units tested so far had a score of over 8 it would mean that the scale would be far too flattering IMHO, not stict enough in other words.
I wouldn't expect anything I have seen ever to deserve 9 out of 10 yet. There is a very long way to go before we see anything even close to perfection in speakers.
Lets see what a state of the art 3-way can do, and I still would not expect state of the 2020 art to deserve 90% personally

When you put it like that you do have a point.

 

[MZKM]

Consider adding:
1. $/Score
2. $/Score_ignore_LFX

This represents $/value, which is very useful when comparing products relative to each other.

How should pricing be formatted? I currently price everything as pairs even if sold individually, except for center channel speakers. Should I keep that, or should I use the price for an individual speaker, even if not sold individually (but in the other case, some companies offer bundle deals, KEF for instance charges $800 MSRP, actual is $700, for a single LS50).

 

[Krunok]

I tend to take the opposite view.
Speakers are by a country mile the worst performing part of a hifi. If one of the small number of modestly priced units tested so far had a score of over 8 it would mean that the scale would be far too flattering IMHO, not stict enough in other words.
I wouldn't expect anything I have seen ever to deserve 9 out of 10 yet. There is a very long way to go before we see anything even close to perfection in speakers.
Lets see what a state of the art 3-way can do, and I still would not expect state of the 2020 art to deserve 90% personally

.. on the other hand - spinorama is all about linearity of direct and reflected sound, so once we reach the linearity limits of our hearing system better mark will stop translating to better sound.

 

[MZKM]

Consider adding:
1. $/Score
2. $/Score_ignore_LFX

This represents $/value, which is very useful when comparing products relative to each other.

Just tried it, and without any real curving, it looks deceptive:

I think a scatter plot like which was done for SINAD and other metrics in another thread is more suitable:

Would a log scaling for price be better?:

 

[pozz]

Wow... I'm thinking unless Amir starts bulking up seriously (or installs a hydraulic lift of some kind in his garage) we're not likely to see any speakers for which 14.7Hz has anything but a simulated response. I'm thinking based on both the values of the KH 80 and the Revel C52... even those Salon 2's might be hard pressed to break into the 9s.

I'm aware of that. But if we can't find a single speaker that will breach the 8 mark than we can start questioning Olive's rating system to be too "strict".

Keep in mind a 10 is reserved for a “perfect” loudspeaker, absolutely no “wiggles” on/off axis, and perfectly controlled directivity both horizontally and vertically.

From what I remember of Toole's book, trained listeners were hard-pressed to rate anything above an 8.

So if the 8s are the true upper boundary, what we have is an indication of the state of our technology, not the flaws of the rating system.

 

[pozz]

By the way, @MZKM I think it would be better to have the price be recorded for one speaker only, and to round to the nearest 10s digit instead of using the x.99 convention.

 

[MZKM]

By the way, @MZKM I think it would be better to have the price be recorded for one speaker only, and to round to the nearest 10s digit instead of using the x.99 convention.

But how would that work with say the LS50, where you can buy them as a pair or buy a single one but for a higher price than 1/2 the pair price?

 

[Krunok]

So if the 8s are the true upper boundary, what we have is an indication of the state of our technology, not the flaws of the rating system.

Or we have a rating systems which evaluates linearity above the capabilities of our hearing system. In other words, even if due to advance in technology we will be able to make more linear speakers still there will be nobody able to hear that advance.

Do you really think we can hear linearity better than say ±1 dB?

 

[MZKM]

Or we have a rating systems which evaluates linearity above the capabilities of our hearing system. In other words, even if due to advance in technology we will be able to make more linear speakers still there will be nobody able to hear that advance.

Do you really think we can hear linearity better than say ±1 dB?

Toole’s research has stated low-Q deviations (0.5dB?) can be audible.

 

[Krunok]

Toole’s research has stated low-Q deviations (0.5dB?) can be audible.

IIRC it was 1dB.

 

[pozz]

Toole has written that: "There is some evidence that we can hear slopes of 0.1dB per octave, which translates to a 1dB tilt from 20Hz to 20kHz. Such a spectral error, if small, is likely to be benign and subject to adaptation: we would simple get used to it."

So questionable in any case, but worth thinking about.

Understanding the scale and its practical ends is more important than suggesting that we change it or question its relevance based on a hazy sense of psychoacoustics. So we might not find a 10, but then designing a speaker with a preference rating of 1 is probably going to be very difficult too.

 

[MZKM]

So we might not find a 10, but then designing a speaker with a preference rating of 1 is probably going to be very difficult too.

Amir shared a paper with me where they used Olive’s formula and got negative scores when testing flat-panel speakers .

 

[pozz]

... a hazy sense of psychoacoustics ....

A small example.

With hearing loss critical bands get wider and the subjective sense of loudness and tone perception shifts. Since this is a natural part of aging you can consider that the numbers quoted for audibility will not change only for absolute thresholds, but also for sensitivity to spectral deviations since more frequencies will fall into a single band.

I haven't read enough to say what happens to sensitivity with sudden hearing loss, or congenital.

So the design goals remain clear.