The point isn't which is the "best" speaker, that's not what Floyd Toole was about, it was which one did the market prefer. This is not about scientific research, it's about market research. Usually when you want to develop a new best seller you include the old best seller and demonstrate it in its best light to see if the market likes your new entry better. Toole wanted to know what was it about a particular entry that caused people to prefer it to some alternative. His conclusions as I understand it is that it has flat FR and creates more lateral reflections which means better dispersion. But these were the goals of AR speakers going back to AR3 in 1960. So to find out whether or not Revel Ultima Salon 1 or 2 is preferred by the market over other speakers, shouldn't the prior top dog be in the pack? Of course that speaker and its successor AR3a hasn't been manufactured for 40 years and the remaining samples are often restored at great effort by collectors. It is not known whether restored versions sound like the original units because virtually all of the original tweeters have deteriorated due to a poor choice of materials. Also improvements to it were possible using more current technology. For example in additon to inventing the dome tweeter and midrange and acoustic suspension woofer Acoustic Research invented ferro fluid cooling to improve tweeter power handling and introduced it with AR 10 pi and AR 11 in the mid 70s. These speakers all struck me as being able to benefit substantially from what is now called "room correction." Therefore to prove a new product likely to be a market preferred model it should be tested that way. Or do you think changes in market preference over time would cause a different result?
It’s All a Preference
- Thread starter mep
- Start date
The point isn't which is the "best" speaker, that's not what Floyd Toole was about, it was which one did the market prefer. This is not about scientific research, it's about market research. Usually when you want to develop a new best seller you include the old best seller and demonstrate it in its best light to see if the market likes your new entry better. Toole wanted to know what was it about a particular entry that caused people to prefer it to some alternative. His conclusions as I understand it is that it has flat FR and creates more lateral reflections which means better dispersion. But these were the goals of AR speakers going back to AR3 in 1960. So to find out whether or not Revel Ultima Salon 1 or 2 is preferred by the market over other speakers, shouldn't the prior top dog be in the pack? Of course that speaker and its successor AR3a hasn't been manufactured for 40 years and the remaining samples are often restored at great effort by collectors. It is not known whether restored versions sound like the original units because virtually all of the original tweeters have deteriorated due to a poor choice of materials. Also improvements to it were possible using more current technology. For example in additon to inventing the dome tweeter and midrange and acoustic suspension woofer Acoustic Research invented ferro fluid cooling to improve tweeter power handling and introduced it with AR 10 pi and AR 11 in the mid 70s. These speakers all struck me as being able to benefit substantially from what is now called "room correction." Therefore to prove a new product likely to be a market preferred model it should be tested that way. Or do you think changes in market preference over time would cause a different result?
So you are saying we could have hired a marketing person and he would have come up with the same findings? He would have known how to isolate variables that corrupt the lab results. He would have known how to device measurements systems of speakers to predict preference? And his work would be impressive enough to appear in ASA and AES journals?
But yes, ultimately good engineering is to satisfy your customers. Someone buying BMW wants great handling. I don't call the engineer who figures out how to deliver that performing market research or not striving for the best.
He started this work while at NRC. There was no best seller before or after. They wanted to cut through the folklore and hundreds of assumptions on what is good sound. They eliminated variables and constantly checked the science to see if they two can be matched. You make it sound like he is a salesman looking for the next bullet to put on the marketing brochure.
It is not quite that. It is not about flatness but smoothness. And it is not just any response but direct and indirect. It happens then that if you have good off-axis response, you can take advantage of reflections to get you perceptually better results.
You mean they were performing market research too? And not creating the best speaker?
No, I don't. If you have measurements of AR-3 we can see how they would fare in this test.
Marketing people often must be guided by scientists and engineers to understand what variables to compare to find out where market preference lies. Then engineers take those variables to create the most ideal product based on those findings. For example, once the favorite smell, texture, color of a sun tan lotion is preferred, scientists and engineers will devise a product that combines them and then test it again against other top selling products to see if the market prefers it. That's how I see Toole's work, market research, what does the market like best, not what would improve accuracy or convincibility. If the market liked juke boxes most, then that's what his end product would have sounded like.
I don't know what ASA would have published, that is a scientific journal but IMO as a former member AES would publish just about anything. As I now see AES it is a "prosumer" organization. I'm no longer impressed with it as a legitimate organization of the highest standards possible. It's become far too commercialized for my liking.
I've seen other products that were results of NRC. For example I sat through about an hour of talk about an "Energy" loudspeaker and when it was finally demonstrated it was of no significance as far as I could tell. It didn't excell at anything, it broke no new ground. I see NRC also as geared to sell Canadian manufactured products whether they are technically superior or not.
"You make it sound like he is a salesman looking for the next bullet to put on the marketing brochure."
That's largely how I see him. Except for his discovery about subwoofer placement which was a real contribution.
"It is not about flatness but smoothness."
Where's the difference?
"And it is not just any response but direct and indirect."
I've already associated that with dispersion at all frequencies including critically at high frequencies. I've already asked about Revel Ultima Salon 2's seemingly poor showing in this regard as indicated by figure 6 in Stereophile's review, comments to the contrary. And I also said I'm not surprised. The AR tweeter is a 3/4 "half hemisphere not recessed. The only criticism of it was the decorative angled molding around the outside of the cabinet that projected forward of the front baffleboard which caused a reflection. Revel's tweeter by contrast is 1", not a full half dome and is recessed in a waveguide focusing its energy to an increasingly greater degree as frequency increases. Show me where JA's measurements or my interpretation of it are wrong. This was further expounded on by him when he made his comments about "airlessness" I referenced in a prior posting.
"You mean they were performing market research too? And not creating the best speaker?"
Their development of wide dispersion midrange and tweeters to optimize their design was playing on a hunch. They didn't do the kind of research Harman did and they probably were as surprised as anyone when they garnered 32.2% of the US speaker market. What you many not know that makes it all the more remarkable is that they treated their distributors like hell, so much so that some of them deliberately sabotaged the demo units in different ways in their showrooms to sell other speakers that made more profit in preference.
"If you have measurements of AR-3 we can see how they would fare in this test."
AR's own speaker measurements are published on The Classic Speaker Pages in their library and are also published in the many reviews they received when they were marketed.
This is a good point to clear up. We see this pop up every time we discuss accuracy etc in speakers. I think we are often talking about different things.
Anyway, you prefer the bass slightly tilted up (personal tase right), I like the midrange slightly emphasised (personal taste right)...that is an example of 'smooth' as opposed to 'flat'. Let's define flat as equal sound pressure level at every frequency, and so by definition that is also smooth (let's define that as in standard english, varying without abrupt change).
So smooth response does not have to be a flat response (tho the reverse I think is a given). It is the sharp abrupt changes in response that are the troublesome ones, and also indicate phase problems etc.
(in a general sense, these waters can get murkier once we have people (unkowingly perhaps) also applying it to the LP or anechoic, or native response of speaker vs room response etc etc.. There are a few 'accepted' at the LP responses that are generally thought to be preferrable by most)
Geez, we do agree on something: the softwareInteresting concept since those who suggested the very same thing eg the importance of the front-enf here have been subjected to endless ridicule.
'Cept how do you "standardize" the front end/software when we all listen to different sources?
Myles: That's two things today that you agree with me. It's been a while since we engaged but I hope you're not going soft on me?
The standardization for recordings would the playback systems and room acoustics through which recordings are monitored and mastered. You can't standardize how the art is made but I don't consider the monitoring environment as part of the art.
So if I understand you, you can have a speaker with a monumental bass where the FR falls off smoothly over the entire audio spectrum until it's very far down, say 10 db by the time it reaches the top octave or two and from what Amir said about Toole's conclusions about smoothness that's a good speaker? It's smooth but not flat so that's okay? And so is the inverse with a very bright treble but a tilt in output downward so that bass is weak? That's a good speaker too by Toole's reckoning? You say an emphesis in the midrange is also okay. By that criteria wouldn't Bose 901 be a good sounding speaker? No highs, no lows, it's Bose? At least that's what they say about it and I think there's more than an element of truth to that.
BTW, I've already made arrangements with a local dealer to audition Revel Ultima Salon 2 and if I decide that they are much better speakers than I already own, I just might buy a pair. There's a pair of the earlier version on e-bay for $7200 right now, about a third of what version 2 would cost new. I'd have to consider that too. I've also said I intend to listen to the MLs again. I intend to keep an open mind about everything and listen to them without any thought about what other people said about them or about my prior impression of ML Summit which was decidedly negative.
It is true that you should ideally measure the loudspeaker in the far-field, and that distance increases as a function of the size of the speaker. We measure at 2 m. which is sufficient for all but the largest speakers. We listen at approximately 10 ft from the loudspeaker so that meets the requirement mentioned by the ML manufacturer. I don't think the peaks and dips you see in the Loudspeaker M measurement are measurement artifacts (diffraction and acoustical interference) related to the measurement distance because they are still visible in the sound power response, and survive 70 spatial averages made around the speaker. These are resonances. Furthermore, the listeners' comments and sound quality ratings correlate well with the acoustical measurements. It's sounds a bit like babble and excuses to me.
The music recordings are a treated as independent variable in the design and the statistical analysis of the test. We typically use at least four different programs when testing loudspeakers. Our approach has been to select tracks on the basis of their ability to reveal problems with the speakers (colorations, bandwidth limitations, dynamics, distortion, spatial artifacts,etc) and they should be generally well recorded so that they don't introduce any complementary biases (e.g. a dull recording making a bright speaker sound neutral and vice versa) to the test. What is considered well-recorded is somewhat subjective but things like low noise, distortion, well-balanced, dynamics, spectrally dense with good bandwidth can be quantified objectively and subjectively.
The listener training software has actually helped identify which music tracks listeners can most easily detect distortions added to music. For spectral distortions, we found that certain tracks are better than others, and that the ones that listeners found most useful tended to produce spectra that was dense and extended in bass and treble. Pink noise was the best. Tracy Chapman was the next best thing to pink noise. Solo instruments and chamber music wasn't very useful ( see the graph showing listener performance as a function of program here. The listener training also helps familiarize the listener with the characteristics of the program, and listeners learn the strengths and weaknesses of programs that are likely accounted for when evaluating components. For example, if program is known to be a bit bright or excessive in bass, listeners will acommodate for those idiosyncrasies. Since we tend to use a multiple comparison method (A/B/C/D or MUSHRA-style method) distortions in the program are constant or common to each component and s tend to perceptually disappear into the background during the evaluation process Listeners tend to pay attention to features that vary as they switch among the different test objects.
We are actually in the process of designing and running some experiments to help us select some new and hopefully more sensitive rogram material at revealing different types of distortions in loudspeakers and other devices. I would appreciate your suggestions for music tracks that you believe to be exceptionally well recorded and revealing of distortions in audio components.
Are you deliberately distorting what I said? Let me know so I can see whether or not it is worth the bother of typing when it has something to do with you would ya?
This is an interesting thread. I think people generally tire of something over time mainly because they want something new. My wife has no interest in my hobby and she is satisfied with casettes and crappy speakers; however she hates bad recordings. The fact that we are even interested in the hobby means we are always predisposed for looking for something new and better even if means deciding that what we have isn't good anymore even though it may very well be....if we could get away from our expectation bias. I mean the audio press relies on this or they would not have something new to review every month that is generally better than anything in the past.
I really don't see anything different in audio than a lot of other things in the "market of ideas". If a scientist comes up with something that we don't agree with we usually say the science is wrong or flawed. I think that is fine for other sientists to do as it is part of the peer review process. I don't put too much faith in the person who has already made up their mind anyway. Just my opinion. And thanks to Floyd and Sean for their work now and with the Canadian Research facility.
I really don't see anything different in audio than a lot of other things in the "market of ideas". If a scientist comes up with something that we don't agree with we usually say the science is wrong or flawed. I think that is fine for other sientists to do as it is part of the peer review process. I don't put too much faith in the person who has already made up their mind anyway. Just my opinion. And thanks to Floyd and Sean for their work now and with the Canadian Research facility.
Agreed. Can you tell me whom are the members of the board of reviewers who carried the peer review of the latest papers we have been debating?
I don't know the names but the Dr. Toole papers were in the journal of AES which I am told has a peer review board. http://www.aes.org/e-lib/browse.cfm?elib=5270
Here are the details: http://www.aes.org/journal/
"The Journal of the Audio Engineering Society — the official publication of the AES — is the only peer-reviewed journal devoted exclusively to audio technology. Published 10 times each year, it is available to all AES members and subscribers."
That said, peer review does not say the results are right but rather, the work is high quality. And the author did not flunk high-school.
"The Journal of the Audio Engineering Society — the official publication of the AES — is the only peer-reviewed journal devoted exclusively to audio technology. Published 10 times each year, it is available to all AES members and subscribers."
That said, peer review does not say the results are right but rather, the work is high quality. And the author did not flunk high-school
.
I don't know who reviewed my papers or Floyd's as the process is normally blind and confidential.
I have reviewed papers for both J. AES and J. ASA on papers related to this subject matter, so someone must think I have expertise in this area.
That said, peer review does not say the results are right but rather, the work is high quality. And the author did not flunk high-school
Amir,
Surely. But in order to be nominated reviewer you are supposed to have published high quality papers in related subjects - and it could be a good clue to find more interesting quality papers on the subject. I have the references of F. Toole in his book, but I am looking for more.
I was not talking about reviewer but the paper/author being reviewed. My team at Microsoft used to handle peer reviews for signal processing journal/conference and what I stated was the criteria. It does indeed take high credentials to do the reviewing.
Just because something is peer reviewed doesn't necessarily give the paper credibility. Not all scientific journals are created equal (read: some journals are much easier to get papers published in than others)! To get an idea, check out each journal's acceptance rate. Ergo, there are A, B and C journals. Not to mention those on the journal's Editorial Board usually get special treatment.
Oh yes, I also reviewed papers for several journals in my field when I was in academics.
Are you objecting or agreeing with me? This is what I said: "That said, peer review does not say the results are right but rather, the work is high quality. And the author did not flunk high-school
"
Are you objecting or agreeing with me? This is what I said: "That said, peer review does not say the results are right but rather, the work is high quality. And the author did not flunk high-school
Both. There are first rate, second rate and third rate journals in every field. So not even necessarily of high quality.
Similar threads
