KEF LS50 Meta Review (Speaker)

[thewas]

Fixed.

I also disagree with the premise, the LS50 has a much less even axial FR than the LS50M.

The differences are easily audible with most music (I own both since they were released and I cannot hear the difference in SINAD above 60 dB etc.), here is also a direct audio recording comparison of both where despite the not very discriminating music samples used the audible difference is really obvious and in accordance to their measurement differences:

 

[gasolin75]

lacking bass, something about the midrange and notice some distortion at the used levels from the white speaker LS50

 

[pablolie]

I learnt to ignore AI when I saw that it gave equal weight to the opinions of an acknowledged expert, and a beginner when I asked about a particular espresso machine. The above looks to have ignored frequency response, among other things. AI does not think, it’s more Artifical YouTuber than Artificial Intelligence.

I didn't show the prompt, I explicitly asked it to exclusively look at measurements and to ignore any subjective listening "impressions".

 

[Leif]

I didn't show the prompt, I explicitly asked it to exclusively look at measurements and to ignore any subjective listening "impressions".

Good to hear, that makes sense. I assume it could not interpret any graphical data as it reads text only. I was surprised to hear a difference in the video, definitely more bass, less treble in the metas. I was listening with LS50 metas, no doubt the speakers or cans used will affect the listener’s impression.

 

[David_M]

The differences are easily audible with most music (I own both since they were released and I cannot hear the difference in SINAD above 60 dB etc.), here is also a direct audio recording comparison of both where despite the not very discriminating music samples used the audible difference is really obvious and in accordance to their measurement differences:

There is absolutely NOTHING similar about these speakers. Claims of sounding identical are patently false.

 

[Arindal]

here is absolutely NOTHING similar about these speakers. Claims of sounding identical are patently false.

Fully agree, absolutely no similarity. If they would not disclose which model is which during the song, I could have been swearing it must be speakers by different manufacturers.

Difficult to judge how they did the recording and how gear as well as room are influencing the result, but the more pronounced coloration of the speaker A (=LS50M) is audible in any case.

 

[thewas]

Difficult to judge how they did the recording and how gear as well as room are influencing the result, but the more pronounced coloration of the speaker A (=LS50M) is audible in any case.

The LS50 non-meta is definitely more colored without EQ, of course an amateur recording in a room is not expedient to show such, but enough to show they sound differently.

 

[pablolie]

Like I said, everybody's entitled to their personal opinion (which is all everybody has posted, no facts, no analysis, just subjective listening impressions - which is fine by me unless presented as indisputable truth). (And posting a youtube quality video as proof is bizarre, sorry. Both because youtube quality is highly unlikely to truly reveal fine differences -unless manipulated- and it also seems to come from a dealer eager to sell upgrades in general, check all the "battle of x vs y".)

All I did is instruct ChaptGPT's most advanced model to study scientific consensus on hear-ability and to just look at measurements, including the graphs on this very website (and yes, 4.5 can read and interpret graphs). Have a great rest of the weekend, and my participation in this topic is over, I am unfollowing it for good.

 

[goat76]

They have made a big mistake in that comparison video. They have moved the whole setup, as can be seen when looking at the wooden pieces on the table, and it wouldn't surprise me if they have moved the microphone too between the recording of the LS50 and the LS50 Meta.

If a comparison like this is made to hear a difference between two loudspeakers, it's very important not to move the microphone at all, and it's important to place the loudspeakers in the same spot and at the same angle towards the microphone.

When it's ensured that the only thing that is changed between the recordings is the swapping of the test subjects, a test like this can be used to at least determine if there is a sound difference between the loudspeakers, but not much more than that.

 

[David_M]

They have made a big mistake in that comparison video. They have moved the whole setup, as can be seen when looking at the wooden pieces on the table, and it wouldn't surprise me if they have moved the microphone too between the recording of the LS50 and the LS50 Meta.

If a comparison like this is made to hear a difference between two loudspeakers, it's very important not to move the microphone at all, and it's important to place the loudspeakers in the same spot and at the same angle towards the microphone.

When it's ensured that the only thing that is changed between the recordings is the swapping of the test subjects, a test like this can be used to at least determine if there is a sound difference between the loudspeakers, but not much more than that.

I see the change in the microphone-speaker distance you are referring to, and you are correct; all should be the same, etc. However, the distance change is a few inches forward, not enough to cause the drastic shift in tone between the speakers in the mid-range. It sounds as though someone attenuated the mid-range EQ settings for some reason.

 

[kyuu]

And posting a youtube quality video as proof is bizarre, sorry.

Agreed. But then so is posting AI "analysis".

 

[pablolie]

Agreed. But then so is posting AI "analysis".

There is zero bizarre about engaging ChatGPT 4.5 in fact checking. Update yourself to the era of prompt programming. Challenge the models.
You'll get far better results from prompt programming and challenging advanced models than sheepishly believing in a YouTube video. Come on.
No one claimed AI is always right. But it is a good tool to establish current consensus (lowest common denominator) ... never to create innovation. Neglect it and become obsolete, though.

PS: I still trust the audio luminaries in here 500% more than I'd do uncurated and uncorrected AI results. But I'd trust subjective opinions and reviews and youtube videos way less.

PS2: You don't get access to the best ChatGPT 4.5 model unless you pay for it. And I repeat - it's nowhere near perfect. And yet pretty amazing. And yawn to everyone that claims I outsource and sheepishly believe it, I challenge it all the time [and you can tell it to not use the new info it gained in any other prompts, and the new user agreement specifically says you can sue the company if it steals your IP.. but you better fast track your patent ideas].

 

[stigger]

Humans have biases. LLMs have biases. When it comes to science, you can trust neither. Only objective measurements and observations are guaranteed to represent the reality.

There is zero bizarre about engaging ChatGPT 4.5 in fact checking.

You should never involve an LLM into fact-checking. You can make logical conclusions only based on objective reality, not its statistical approximation.

 

[pablolie]

Humans have biases. LLMs have biases. When it comes to science, you can trust neither. Only objective measurements and observations are guaranteed to represent the reality.

You should never involve an LLM into fact-checking. You can make logical conclusions only based on objective reality, not its statistical approximation.

Disagree with that premise. The best LLM models are *provenly* very effective in summarizing stuff. Without that capability they'd be unable to make a single $. What they can never do is build on top of that.

 

[thewas]

It is telling how measurements, recordings and experiences of people are ignored while the known flawed summary of a LMM based on undefined data is placed as a supposedly equal counterargument.

For people who really are interested in testing how audible the differences and I had made some filters with which someone can enable their differences with a switchable EQ:

Filter  1: ON  PK       Fc   55.30 Hz  Gain  -1.00 dB  Q  3.000
Filter  2: ON  PK       Fc   140.0 Hz  Gain  -1.00 dB  Q  2.000
Filter  3: ON  PK       Fc   450.0 Hz  Gain   1.50 dB  Q  1.400
Filter  4: ON  PK       Fc    1500 Hz  Gain   2.00 dB  Q  3.000
Filter  5: ON  PK       Fc    2600 Hz  Gain  -3.00 dB  Q  1.700

If someone doesn't trust my measurements they of course can feel free to do the same with any measurement set of Spinorama.org (luckily both are some of most measured loudspeakers) and even use listening window, early reflections or sound power measurements if they say they are listening at a very short, medium or very large distance (hint: all are different enough that with most wide spectrum music the differences are easily discernible).

 

[goat76]

I see the change in the microphone-speaker distance you are referring to, and you are correct; all should be the same, etc. However, the distance change is a few inches forward, not enough to cause the drastic shift in tone between the speakers in the mid-range. It sounds as though someone attenuated the mid-range EQ settings for some reason.

The entire setup is also shifted approximately 6-8 centimeters to the side, as can be seen by examining the different wooden pieces on the table. Additionally, I couldn't find any information on how they captured the audio, so it's impossible to judge based solely on the camera position, which may or may not have been used for the audio capture.

 

[MAB]

Like I said, everybody's entitled to their personal opinion (which is all everybody has posted, no facts, no analysis, just subjective listening impressions - which is fine by me unless presented as indisputable truth). (And posting a youtube quality video as proof is bizarre, sorry. Both because youtube quality is highly unlikely to truly reveal fine differences -unless manipulated- and it also seems to come from a dealer eager to sell upgrades in general, check all the "battle of x vs y".)

All I did is instruct ChaptGPT's most advanced model to study scientific consensus on hear-ability and to just look at measurements, including the graphs on this very website (and yes, 4.5 can read and interpret graphs). Have a great rest of the weekend, and my participation in this topic is over, I am unfollowing it for good.

It's just that they do measure differently, or are the measurements no longer important?

Ironic that ChatGPT is a sycophantic distillation of personal opinion. Perhaps stop with the ChatGPT, it's not actually your friend.

You've made this argument before, ignoring measurements. People have actually provided you with measurements and analysis, your personal opinions seem to be overriding. Tedious to hear it again, now bolstered by an AI partner.

 

[pablolie]

It's just that they do measure differently, or are the measurements no longer important?

Ironic that ChatGPT is a sycophantic distillation of personal opinion. Perhaps stop with the ChatGPT, it's not actually your friend.

You've made this argument before, ignoring measurements. People have actually provided you with measurements and analysis, your personal opinions seem to be overriding. Tedious to hear it again, now bolstered by an AI partner.

I never once ignored measurements, that's a blatant lie [pointer to another message below]. And to me, those who claim they can reliably hear a <2dB difference over a limited FR and say that represents a "huge difference" are just as laughable as those who claim their $20k speaker cable makes a huge difference, sorry. And in this case conveniently ignore the Meta has a larger deviation from "ideal" - which as I have stated repeatedly didn't bother me the least when I listened to them. My point was never ever once the previous generation was "better", just that given the fact whatever small differences I thought I heard never improved or negatively impacted my listening experience, I decided to keep the speaker that came in my fav color.

KEF LS50 Meta Review (Speaker)

Thanks for summing it up perfectly. Audio discussions sometimes remind me of the joke about Leibniz and Democritus arguing about monads vs atoms. :-D Aren't the bitterest battles fought over the smallest stakes? I think I just mangled Sayre's Law: "In any dispute the intensity of feeling is...

audiosciencereview.com

Takeaway:

 

[MAB]

I never once ignored measurements, that's a blatant lie. And to me, those who claim they can reliably hear a 1dB difference and say that represents a "huge difference" are just as laughable as those who claim their $20k speaker cable makes a huge difference, sorry.

KEF LS50 Meta Review (Speaker)

Thanks for summing it up perfectly. Audio discussions sometimes remind me of the joke about Leibniz and Democritus arguing about monads vs atoms. :-D Aren't the bitterest battles fought over the smallest stakes? I think I just mangled Sayre's Law: "In any dispute the intensity of feeling is...

audiosciencereview.com

Takeaway:

View attachment 467481

That's not 1 dB.
But this was already discussed.
They are nice speakers, there are other nice speakers too.
Among nice speakers, it is useful to be able to compare and discuss actual differences. These are actually different, in a region that we can actually hear differences.

 

[pablolie]

That's not 1 dB.
But this was already discussed.
They are nice speakers, there are other nice speakers too.
Among nice speakers, it is useful to be able to compare and discuss actual differences. These are actually different, in a region that we can actually hear differences.

1. You didn't apologize for claiming I ignored measurements.
2. Now you argue about 1dB or 2dB... tomato tomatoe. That's never been the huge issue - the *character* of the speakers is remarkably similar in peaks or dips over a limited FR. Plus take into account Kippel states their measurements are accurate +/- 1dB, so arguing whether it's 1 or 2dB doesn't make any sense to begin with.
3. Also can't acknowledge that, if such small things make a huge difference, the OG should be regarded as "better" since the Meta has the bigger deviations from "ideal" over the presence region.

Weak response, I consider you a much better participant than that. I respect both you and thewas a lot for your contributions... outside this topic. If you were to say the presence region is where a more analytically tuned speaker can sound a bit harsh, and you prefer a slightly softer delivery there for home listening, it'd be perfectly fine.

PS: As to the strengths and weaknesses of current AI models, there's a dedicated discussion there. Take your AI-is-useless argument there and see how well it's taken. :-D