Some More Evidence that Kids (American and Japanese) Prefer Good Sound

[microstrip]

I do 99% of my listening alone as well, Micro. A narrow projection in a dipole strikes me as odd, but what I really just don't get about the ML thing is the tonality. To me, they always sounded just like those Harman charts look -- ragged, uneven and rolled off rapidly at the bottom. But maybe I never found an amp that made them work. I definitely don't get it. Occasionally a customer would fall in love with them, and I never quite got why. Probably just a personal taste thing. I'm a subjectivist.

Tim

Tim,

Do you have experience with the Quad ESL63's?

 

[Phelonious Ponk]

Tim,

Do you have experience with the Quad ESL63's?

I heard some Quads years ago, I don't know what the model number was. I was impressed at the time.

Tim

 

[NorthStar]

---Godd (good) Science is to always question the tests and the results, and improve on them.

* Synergy. ...Key word (or System Matching; good one).

 

[tonmeister2008]

Of course I do think that marketing and advertising matter. That’s why I said that if Harman is building speakers that are technically superior to the competition and with Harman’s ability to fund marketing and ad campaigns, you should be able to rule the speaker world in terms of sales. Who could afford to outspend Harman in the advertising and marketing realm?

As recent as 2 years ago, Bose spent about 20x more money on advertising than any Harman brand. We have only recently begun spending money on advertising -- including TV spots -- Dr. Harman believed in word-of-mouth, and not a big believer in expensive advertising campaigns. There is finally the realization that brand recognition and equity are not unrelated to advertising. if we want to continue our large marketshare in the branded audio automotive space (which accounts for ~ 70% of our sales) we need to spend money to harvest brand equity. Our automotive customers demand this.

 

[tonmeister2008]

Fortunately, audiophiles are massed at the extreme end of the Bell Curve distribution.

But which end?

 

[tonmeister2008]

Let me clarify this first... The biases that I see are 100% with the process of putting the test together, not with the evaluators, and consequently with the people behind putting the test together. The biases involve the software, hardware and the room.

Starting with the software biases I see, the musical material appears "light" and not very demanding to me, but I have to admit I am not familiar with it. I am only guessing there was voice, a guitar and perhaps a couple of other strings and maybe some light drumming involved. However, having lived with the similar $3K ML SL-3 in the past and also owning $600 Monitor Audio mini-monitors in another system, I can safely guess that, if one were to put on much more demanding material on - like Mahler's 2nd - I would hazard a guess the cheap dynamic speakers in the test would fall apart and the ML would hold well. I get really beautiful sounds with quartets from my Monitor Audios and they just [naturally] suck at large orchestral.

On to the hardware bias... It's a very common mistake to assume that any sufficiently powerful amp can properly drive difficult loads as an ML - flat out: NO; I can't stress this enough. I have posted in the past that I was about to purchase the ML Summit X last September (a more difficult load than my Odysseys), had a custom configuration quoted for me, only to kill the deal at the very last audition. Why? Because even the very capable, close-to-state-of-the-art and one of the top amps in the market Spectral 360 Series II monos that I use sounded harsh up top and would eventually shut down, and they are one of the few amps rated at 2ohms. In case anyone thinks that I am simply making a statement to win an argument, here's a PM I sent Myles back then, so he can corroborate the truth of my claims:

So I am thinking twice about the X - not a good match with the Spectrals, it seems. I took another listen today, and the sound is aggressive from the midrange up, tilting doesn't really help - it's not a tonal balance issue - and the amps shut down with high-frequency content like mass brass on Stravinsky's Rite of Spring, at high enough (but not ear splitting) volumes. Overall, the X is a more resolving speaker than the Odysseys - crisper - but the system right now is so beautiful that I wonder why do I need to break it. If nothing else, I might look at rebuilding the average-quality crossover.

Next, the room - am I right that the room was unechoic? A dipole is not meant to have its back wave absorbed - this is a fundamental flaw.



There are a couple of points here. The test simply shows that most people like dynamic speakers over panel ones; at least with the material used. _We know that_. That's why out of the 600 speaker manufacturers out there, there are only a dozen specializing in panels, and size differences (the visual aspect) is not enough to account for this disparity. The test should have used an expensive dynamic speaker as the poster child. For what is worth, I stated the other day that my own references are primarily the Magico Q3 and Q5, with the ML CLX close behind. But they come from two completely different schools of thought, so it's unfair to directly compare them, unless the room, software and hardware are equally optimized for all subjects under test - an extremely difficult feat. I just don't see this requisite unilateral optimization in the test we are discussing.

Finally, if I haven't driven this point hard enough, it is hard to drive a panel correctly, and it's equally hard for many listeners to like the panel sound. I'll save all the details for another thread, but to give a common example, panels can render unrealistically large images - that's a big turnoff for most people. But, alas, this is not necessarily a function of the speaker alone; I have had a number of amps inhouse and only the current Spectrals can make my ML's sound small when the program calls for it, likewise large when appropriate. It's not by happenstance that serious ML setups (and frankly, all Magnepan and Analysis Audio I have heard) run these speakers with very capable (and unfortunately, expensive) amplifiers.

If the A/B test wanted to point out that inexpensive dynamic speakers can be preferable to unoptimized panels with light material, it succeeded. But it's biased in my view, and simply points out common preferences.

All four speakers are subjected to the same software and hardware so any biases are constant. The music was not a statistical significant factor, and sufficiently neutral that the most accurate speakers were the highest rated, and the least accurate speakers were the lowest rated. The music included drums, bass guitar, vocals, piano and other guitars and the spectral analysis does not support the claim that it was "lightweight". There is statistical evidence showing the material is best a revealing audible differences among loudspeakers and added resonances. So, explain to me how the music biased the results?

The amplifier was a Mark Levinson 532H, with adequate power to drive the speakers at the listening volumes used in this test without clipping.

Given the objective measurements of the loudspeakers, it seems highly unlikely that any music program would make Speaker D sound natural and accurate given the presence of multiple resonances in the measurements. Explain to me how a music signal would not excite the resonances clearly apparent in the measurements.

The listening room is semi-reflective: the walls and ceiling are reflective so the back waves from the speakers were reflected -- not absorbed.

I don't think the test shows evidence that people prefer dynamics over panel speakers. That wasn't the point of the test and there aren't enough sample points to come to that conclusion. (The point was to see if there is agreement in loudspeaker preference among trained and untrained listeners including Japanese and American college students.

I can find examples of dynamic speakers that would measure and rate lower than speaker D, although it would be quite a challenge. And I could probably find some electrostatic speakers (like the old Quad) that would outperform some dynamic models. It all comes down to how competent the engineering is done, and if you have access to advanced subjective and objective measurements you can easily optimize and verify the sound quality of the design.

 

[Bjorn]

Quad Reference ESL-2805 have been reviewed and measured in the May edition of Stereophile. Where they really excell is in low distiortion and step response.
Here are two dynamic speakers that measures worse then I've ever seen:
http://www.soundstagemagazine.com/measurements/zucable_druid/
http://www.soundstagenetwork.com/measurements/speakers/gemme_vflex_katana/

 

[Tonepub]

I think Dr Olive and his team are on to something, those $500 Infinities were pretty damn good. If they wouldn't have said something, I would have easily guessed that they cost 2 -3 times more easily. That's the advantage the big companies have over the "fred in the shed" speaker mfrs. Their economies of scale allow them to produce some pretty awesome stuff for very reasonable $$. Focal has done some pretty amazing things on the low end of the price spectrum too. And Paradigm.

 

[microstrip]

I heard some Quads years ago, I don't know what the model number was. I was impressed at the time.

Tim

Tim,

The ESL57 are still considered a reference concerning timbre, stereo imaging and natural sounding.
Would you buy them looking at the frequency response? It seems to fit your description - ragged, uneven and rolled off rapidly at the bottom (and top, btw).

The ESL63 is one the speakers I could happily live with - they were my company for almost 20 years (with a few interruptions, I must say).

 

[Bjorn]

Sean:
Do you expect another results when it comes to what speaker response people prefer if they are mostly listening to one music genre and where most recordings are bad?

Are there researches from other then Harman Int. about speaker preference that correlates with yours? And if so, which?

B&K studies in the 70s showed that a response needed to gradually fall with 3dB from bass to treble in the room to sound correct tonality. Is this the response you also today reckon as being the most correct one?
Would appreciate if you could tell more in detail what you consider an accurate speaker to measure like.

 

[NorthStar]

---I don't know how many of you are familiar with BOSE table clock radios,
with integrated CD player; but they are the best sounding ones of that type.

I know; my Mum has one, my next door neighbor has one, ...
and they are surprisingly great sounding. :b
Think 'bout it: you got a tuner, a preamp, an amp, a CD player, speakers (stereo), integrated subwoofer, ...
you can put it anywhere, don't take much place, nice style, remote controlled, and for only $500.

 

[tonmeister2008]

There's been some reporting of this paper in the press by CEA and HomeTheaterReview.com today by Jerry Del Colliano who believes I should prove that High Res is preferred to CD. I'm not sure why that responsibility falls on my shoulders.

Let' assume that one could prove this: would it matter to the music industry given that they care about profit than sound quality? You would have to make a business case for high res to make them wake up from their 20+year nap. Then again, their current approach doesn't seem to be working either.

 

[amirm]

The hometheater article seems rather strange. It is a grumpy piece but for what reason? He should be happy that a differentiation was found between CD and MP3. Not to be demanding that the same be shown for 24/192. I suspect that comparison would not be remotely as differentiated as MP3 and CD were.

He also makes technical mistakes which I expect to see in non-technical outlets, not what he writes about. He says, "The sound from a compact disc has much more resolution and is much less compressed..." . There is no resolution difference between MP3 and CD. Both are 16/44.1 Khz. And CD is not compressed.

 

[tonmeister2008]

Sean:
Do you expect another results when it comes to what speaker response people prefer if they are mostly listening to one music genre and where most recordings are bad?

Are there researches from other then Harman Int. about speaker preference that correlates with yours? And if so, which?

B&K studies in the 70s showed that a response needed to gradually fall with 3dB from bass to treble in the room to sound correct tonality. Is this the response you also today reckon as being the most correct one?
Would appreciate if you could tell more in detail what you consider an accurate speaker to measure like.

My results correlate very well with the research done by Toole et al at the National Research Council in Canada before we joined Harman. Many loudspeaker companies have read this research (it's published in the J.AES) and some have incorporated it into loudspeaker engineering guidelines.

The research findings would also seem to be agreement with loudspeaker-related papers published in the J. AES and J. ASA by Bech, Gabrielsson, and in particular, Wolfgang Klippel. I've found no papers that showed evidence that speakers having linear distortions in their measured frequency response are preferred. Most of the controversy lies in where and how to best measure the loudspeaker (anechoic, in-room, direct on-axis versus sound power, 1/3 octave vs. 1/20-th octave resolution,etc), and the exact perceptual role that the direct and reflected sounds have on listeners' loudspeaker preferences. The acoustics of the listening room play a role here. The reflected sounds produced by the loudspeaker play a more important role as you move into more reflective listening spaces. We have shown evidence that both the direct and the reflected sounds play an important role in typical semi-reflective listening rooms.

Regarding the influence of recordings on loudspeaker preferences: Well, if you believe Myles, these results ARE based on "bad" recordings, even though the most accurate speakers are preferred with these "bad" recordings But joking aside, it's difficult to design a speaker for "bad" recordings because recordings can be bad in so many different ways: poor spectral balance, band-limited, poor instrument balance, spatial attributes, dynamic range, noise, distortion, etc. In my view, loudspeaker companies should design the speakers for good recordings and let the user tweak the recordings with their tone controls if they wish when they encounter bad recordings. Why encourage the recording industry to make bad recordings by designing audio equipment catering to their bad habits?

The B&K study you refer to talks about in-room measurements that include both the direct and reflected sounds. This loudspeaker in-room measurement would approximate our calculated anechoic sound power response or what we call the "predicted in-room response" (not including the room interaction effects below 300 Hz). If you look at our Room Correction study (slide 24) it shows that the preferred in-room target response closely approximates the B&K recommendation. However, we advocate you still need comprehensive anechoic measurements of the loudspeaker before blindly equalizing in a room because the in-room measurements cannot easily separate out the qualities of the on and off-axis sound components produced by the loudspeaker. If the speaker has good direct sound and poor reflected sounds (or vice versa) blind in-room equalization may fix one component at the expense of the other. So you need to start off with a well-designed loudspeaker before attempting to correct it in a room with EQ.

 

[tonmeister2008]

The hometheater article seems rather strange. It is a grumpy piece but for what reason? He should be happy that a differentiation was found between CD and MP3. Not to be demanding that the same be shown for 24/192. I suspect that comparison would not be remotely as differentiated as MP3 and CD were.

He also makes technical mistakes which I expect to see in non-technical outlets, not what he writes about. He says, "The sound from a compact disc has much more resolution and is much less compressed..." . There is no resolution difference between MP3 and CD. Both are 16/44.1 Khz. And CD is not compressed.

Yes, a bit grumpy as he is trying to use the research to advocate to record executives to support high res Blue-ray music. I agree that it would be a challenge to design a listening test to show people prefer high res over CD-quality. To my knowledge, no one has done this yet.

 

[tonmeister2008]

The hometheater article seems rather strange. It is a grumpy piece but for what reason? He should be happy that a differentiation was found between CD and MP3. Not to be demanding that the same be shown for 24/192. I suspect that comparison would not be remotely as differentiated as MP3 and CD were.

He also makes technical mistakes which I expect to see in non-technical outlets, not what he writes about. He says, "The sound from a compact disc has much more resolution and is much less compressed..." . There is no resolution difference between MP3 and CD. Both are 16/44.1 Khz. And CD is not compressed.

Speaking of High Res audio, on the weekend I downloaded some free high res samples here which are available as 24-bit 96/192k Flac files, as well as on Blu-ray. They sound pretty good at home on my current modest audio system.

But a big part of the sound quality is the recording engineers' choice of high quality microphones (all wide bandwidth, linear, low noise DPA omni-directional microphones) and common-sense recording techniques with no apparent use of dynamic compression or limiting.

 

[Bjorn]

Yes, a bit grumpy as he is trying to use the research to advocate to record executives to support high res Blue-ray music. I agree that it would be a challenge to design a listening test to show people prefer high res over CD-quality. To my knowledge, no one has done this yet.

http://theaudiocritic.com/plog/index.php?op=ViewArticle&articleId=4&blogId=1

Thanks for the reply to my questions. I feel I could go on with more....

 

[Phelonious Ponk]

Please go on. Whenever anyone asks Sean good questions, I learn new things. By the way, that Meyer and Moran article, which has been questioned about every way it can be and is highly controversial, shows that listeners can't hear a difference between hi-res and 16/44.1

Tim

 

[amirm]

Speaking of High Res audio, on the weekend I downloaded some free high res samples here which are available as 24-bit 96/192k Flac files, as well as on Blu-ray. They sound pretty good at home on my current modest audio system.

But a big part of the sound quality is the recording engineers' choice of high quality microphones (all wide bandwidth, linear, low noise DPA omni-directional microphones) and common-sense recording techniques with no apparent use of dynamic compression or limiting.

That's exactly it. Huge amount of improvement comes from better recording practices and no being forced to increasing the loudness for CD/MP3 release. For me that is the real reason to adopt high resolution audio. Any improvement coming from 24 bits and higher sampling rate is gravy .

 

[NorthStar]

There's been some reporting of this paper in the press by CEA and HomeTheaterReview.com today by Jerry Del Colliano who believes I should prove that High Res is preferred to CD. I'm not sure why that responsibility falls on my shoulders.

Let' assume that one could prove this: would it matter to the music industry given that they care about profit than sound quality? You would have to make a business case for high res to make them wake up from their 20+year nap. Then again, their current approach doesn't seem to be working either.

-------That guy, Colliano, is he credible, reliable, honest, and solid?

For me, personally, he has been buried a long time ago. And same for his site and most of his reviewers staff.
If I was you, Sean, I would simply ignore him. But that's me, and everything I stand for. :b

Cheers,
Bob

P.S. By the way, high res audio (96/24, etc.) sounds much better than CD (44/16) to my ears. :b