If the only audible difference left between different digital systems is with the speakers, personally, I'd like to know what's objectively the best speaker. So, objectively speaking, what's the most accurate, or honest, speaker made? Which type of speaker design is the best overall? Is it even possible to build a 'transparent' speaker? How can you properly compare two sets of speakers?
By the way, personally, I'd vote for electrostatic speakers.
If the only audible difference left between different digital systems is with the speakers, personally, I'd like to know what's objectively the best speaker.
There is no 'one' best, and the question is even more unanswerable since you havent' specified ANY performance criteria -- how loud they'll have to play, what frequency range they must cover, any constraints on price, appearance, location, etc.
There are, however, loudspeakers that 'measure well' according to the parameters that Floyd Toole and Sean Olive of the NRC and Harman, and others, have found to correlate well with trained listener preference for 'good sound'. Finding these depends on the availbility of comprehensive published measurements and your ability to interpret them. About the best 'clearinghouse' I know of right now that begins to approach this is
http://www.soundstageav.com/speakermeasurements.html (http://www.soundstageav.com/speakermeasurements.html)
which lists only reviews that are accompanied by NRC measurements
So, objectively speaking, what's the most accurate, or honest, speaker made? Which type of speaker design is the best overall?
see above.
Is it even possible to build a 'transparent' speaker?
Not likely. That whole pesky 'electromechanical transduction' thing is inherently a 'major' source of distortion, compared to other parts of the signal chain.
How can you properly compare two sets of speakers?
Double blind, as usual. But it's not something the average consumer is set up to do. Some loudspeaker manufacturers (like the Harman group) do perform them on, though.
Typically consumers listen in stores and buy what sounds best. A minority auditions several pairs at home, usually serially. Others look up reviews and bench tests. I've argued elsewhere that sighted comparison of loudspeakers is so fraught with bias that one would do at least as well, if not better, by just buying based on reported good measurements -- assuming one is buying based ONLY on 'accuracy' of sound, with NO other considerations, a condition that is rarely if ever met.
Needless to say, this idea is often met with derision from those who insist that only their ears can tell them what is 'the best'. (I agree with them, but point out that they aren't really using 'only' their ears, so they still don't what what their ears alone say are 'best')
By the way, personally, I'd vote for electrostatic speakers.
If you really want to get into these questions in-depth, buy Floyd Toole's book "Sound Reproduction".
Klein+Hummel (http://www.klein-hummel.com/klein-hummel/icm_en.nsf/root/home) builds some nice Pro speakers like the O 300 Series (http://www.klein-hummel.com/klein-hummel/icm_en.nsf/root/prof-monitoring_studio-monitors_nearfield-monitors_O300).
If the only audible difference left between different digital systems is with the speakers, personally, I'd like to know what's objectively the best speaker. So, objectively speaking, what's the most accurate, or honest, speaker made? Which type of speaker design is the best overall? Is it even possible to build a 'transparent' speaker? How can you properly compare two sets of speakers?
By the way, personally, I'd vote for electrostatic speakers
Run a search on the web for Sean Olive and his AES publications. He has done a number of DBT experiments in regard to certain speaker setups and discusses them frequently in his personal blog. It's pretty interesting research. He has an account here on HA and has spoke about this in previous posts as well.
http://seanolive.blogspot.com/ (http://seanolive.blogspot.com/)
It's difficult to know which speaker would objectively qualify as best as each one has to suit its purpose (small rooms, large halls, deep bass, etc). One of the most technically elegant, IMO, is the Quad ESL63 electrostatic. A neat example of thinking outside the box in every sense. I also thought it sounded superb but I doubt it would be best in all circumstances.
There is no 'one' best, and the question is even more unanswerable since you havent' specified ANY performance criteria -- how loud they'll have to play, what frequency range they must cover, any constraints on price, appearance, location, etc.
Is a typical listening / living room environment, and realistic volume levels, specific enough in this case? Please forgo any consideration of appearance, price, etc.
If you really want to get into these questions in-depth, buy Floyd Toole's book "Sound Reproduction".
Is perhaps there some particularly useful conclusions that you think this book comes to?
It's hard to believe that someone on this forum would even ask this question.
However, objectively speaking, there is a fastest car, most popular sport, most expensive food and even the most often chosen favorite color. I wasn't asking for a subjective favorite or opinions, I was simply wondering how to objectively evaluate speakers and speaker designs for accuracy, is that really so hard to believe?
This (http://www.eminent-tech.com/RWbrochure.htm) is probably the BEST subwoofer!
It requires no equalization to achieve flat response to below 1Hz.
I think it costs $13,000 USD plus "architectural installation", and you have to provide your own amp.
Double blind, as usual. But it's not something the average consumer is set up to do.
So how can an average consumer correctly compare two sets of speakers, or let me put it to you this way, how would you go about selecting between two pairs of speakers if your criteria was simply accuracy?
Run a search on the web for Sean Olive and his AES publications. He has done a number of DBT experiments in regard to certain speaker setups and discusses them frequently in his personal blog. It's pretty interesting research. He has an account here on HA and has spoke about this in previous posts as well.
http://seanolive.blogspot.com/ (http://seanolive.blogspot.com/)
This thread Blind tests on speakers (http://www.hydrogenaudio.org/forums/index.php?showtopic=69350) may help answer the question(s) asked by the OP. krapple pointed me to Sean Olive. Sean Olive has been part of research projects where they have blind tested speakers. He has a paper on his website that presents, among others, this conclusion.
4) There were clear correlations between listeners’
loudspeaker preferences and a set of acoustic anechoic
measurements. The most preferred loudspeakers had the
smoothest, flattest, and most extended frequency
responses maintained uniformly off axis.
However, objectively speaking, there is a fastest car, most popular sport, most expensive food and even the most often chosen favorite color. I wasn't asking for a subjective favorite or opinions, I was simply wondering how to objectively evaluate speakers and speaker designs for accuracy, is that really so hard to believe?
I assume those in the "audiophile community" have opinions about which speakers are the best, and I assume that there are a few considered to be "reference standards". (Personally, I don't keep-up on that stuff.) Speaker evaluation does require listening and human perception (and perhaps opinion).
And, the "
fastest car (http://en.wikipedia.org/wiki/ThrustSSC)" may not be the "best car".
I'm sure it's possible to figure out what is objectively the most neutral speaker - but lots of people wouldn't pick a neutral speaker as the best sounding speaker. What is the best sound will always remain subjective. So I would agree with the poster that argued that it's not possible to say what is the best speaker for you in the same way that you and I probably will disagree which car is the best - because we have different criteria both in what we need in a car, and what sound we like the best.
I'm sure it's possible to figure out what is objectively the most neutral speaker - but lots of people wouldn't pick a neutral speaker as the best sounding speaker. What is the best sound will always remain subjective. So I would agree with the poster that argued that it's not possible to say what is the best speaker for you in the same way that you and I probably will disagree which car is the best - because we have different criteria both in what we need in a car, and what sound we like the best.
Yah, ok, but I really wasn't asking about the 'best' speaker for me, and yes, of course I completely agree that personal sound preference is a subjective thing. However, what I wanted to know was if there was something like the most accurate speaker or speaker design. I'd guess you could substitute the word accurate for say, transparent or neutral, but I wouldn't be sure as I'm guessing that none of these terms is objectively meaningful in a scientific test. So, really, my question remains, is there such a thing as a scientific, or objective reference, or test, for speaker performance? Just exactly how can an average consumer select the most accurate from among several speakers that all sound good to that person?
It seems to me that speaker accuracy should be important around here, no?
> If the only audible difference left between different digital systems is with the speakers,
It is the loudspeaker and room that have the biggest influence.
> personally, I'd like to know what's objectively the best speaker.
Equipment like CD players and amplifiers have unambiguous transfer functions which enables accuracy to be both defined and assessed in a straightforward manner. This is not the case for a loudspeaker. The accuracy of the direct sound of a loudspeaker can be assessed in a similar manner to an amplifier but the indirect sound cannot. There is no correct directivity for a loudspeaker although there are good and bad choices.
Double blind, as usual. But it's not something the average consumer is set up to do.
So how can an average consumer correctly compare two sets of speakers, or let me put it to you this way, how would you go about selecting between two pairs of speakers if your criteria was simply accuracy?
The average consumer can't ever objectively compare two speakers by walking into a store and trying them. It's even more frustrating when you bring speakers home and they sound nothing like they did in the shop.
However, if you assume the following.
1. You are buying a speaker based purely on measurements.
2. You support Sean Olive's conclusion that "4) There were clear correlations between listeners’ loudspeaker preferences and a set of acoustic anechoic
measurements. The most preferred loudspeakers had the smoothest, flattest, and most extended frequency responses maintained uniformly off axis."
Then just go onto any website with speaker measurements (measured in a anechoic room) and pick the one that fits the criteria Sean Olive has stated in the assumption above. If that speaker happens to be made out of potting clay mixed with smelly cow dung, is the size of a small truck, and makes people vomit from garishness, we'll that's not your concern because you've picked the best speakers given the above criteria.
My point is, blind tests have demonstrated that we do have some criteria when deciding what measurements we want our speakers to have. As for everything else, you can't blind test that stuff.
Personally, I think the paradigm Studios seem to fit the bill nicely. The seem to have a reasonably flat FR Paradigm Studio 100 (http://www.soundstagemagazine.com/measurements/paradigm_studio100_v3/), they have no offensive smell, they have a decent size (your spouse may not agree), and they not visually offensive in an obvious way (your spouse may not agree).
The lower model Paradigm Studios (60 and 20) are also less offensive to the wallet than the Studio 100.
There's a nice explanation of frequency response numbers here (http://forum.ecoustics.com/bbs/messages/34579/131062.html). This is the most commonly used objective measurement for speakers.
I'm sure it's possible to figure out what is objectively the most neutral speaker - but lots of people wouldn't pick a neutral speaker as the best sounding speaker.
Lots more probably
would, if the Olive/ Toole data are to be believed.
That's the crucial point of their research...measured accuracy DOES correlate to listener preference -- people generally
do tend to prefer accurate ('neutral') loudspeakers....IF they are allowed to choose speakers based on SOUND ALONE (without other biases intruding). This explains why the msot accurate speakers aren't necessarily the market leaders. People's judgement of sound quality can be highly colored by non-audible biasing factors, like price, brand reputation, finish, or reviews they've read.
There's a nice explanation of frequency response numbers here (http://forum.ecoustics.com/bbs/messages/34579/131062.html). This is the most commonly used objective measurement for speakers.
and here's explanations for the NRC suite of measurements
http://www.soundstagemagazine.com/measurem...oudspeakers.htm (http://www.soundstagemagazine.com/measurements/test_loudspeakers.htm)
10 years ago I did a DIY speaker installation in my car. Visaton, the supplier of my front speakers offered two tweeters rated for exactly the same frequency response +/- 0.15db, but one had a cone out of metal and the other out of some textile fabric. The salesman recommended the aluminium version for listeners preferring metal and electronic music and the other one for classical and jazz. He had a switchboard for presentation purposes and the difference was instantly noticeable, even when I closed my eyes and turned my back.
What kind of objective measure would describe such differences? It would be probably part of the distortion metrics, but those usually only cover the overall amount and not their "color".
I'm sure it's possible to figure out what is objectively the most neutral speaker - but lots of people wouldn't pick a neutral speaker as the best sounding speaker. What is the best sound will always remain subjective. So I would agree with the poster that argued that it's not possible to say what is the best speaker for you in the same way that you and I probably will disagree which car is the best - because we have different criteria both in what we need in a car, and what sound we like the best.
Yah, ok, but I really wasn't asking about the 'best' speaker for me, and yes, of course I completely agree that personal sound preference is a subjective thing. However, what I wanted to know was if there was something like the most accurate speaker or speaker design. I'd guess you could substitute the word accurate for say, transparent or neutral, but I wouldn't be sure as I'm guessing that none of these terms is objectively meaningful in a scientific test. So, really, my question remains, is there such a thing as a scientific, or objective reference, or test, for speaker performance? Just exactly how can an average consumer select the most accurate from among several speakers that all sound good to that person?
It seems to me that speaker accuracy should be important around here, no?
Sure, a loudspeaker with a flat frequency response and thus able to play back sound that is closest to the original signal would be what you would call most neutral. That speaker would probably be a very good starting point in your quest towards good loudspeakers Finding that speaker shouldn't be very difficult to do by measuring - if you have a decently dead room lying around
10 years ago I did a DIY speaker installation in my car. Visaton, the supplier of my front speakers offered two tweeters rated for exactly the same frequency response +/- 0.15db,
<snip>
What kind of objective measure would describe such differences?
How many frequencies were used to determine the "frequency response"? Two? Twenty? A thousand?
EDIT: Removed claim I have yet to find the citation I'm looking for.
Isn't a sweep used for such purpose? This could be plotted as a continuous function without a limited number of intermediate steps.
EDIT:
Because a some people here seem to be such fans of Sean Olive I googled him. In his last article (http://seanolive.blogspot.com/2009/01/what-loudspeaker-specifications-are.html) about speaker quality metrics he's confirming what I conjectured in my upper post about distortion and a lack of usable metrics, regarding their relationship to a speaker's sound. So the observed phenomenon seems perfectly plausible without the need of a frequency response fraud conspiracy theory about a manufacturer you probably don't even know.
3. The relationship between perception and measurement of nonlinear distortions is less well understood and needs further research. Popular specifications like Total Harmonic Distortion (THD) and Intermodulation Distortion (IM) do not accurately reflect the distortion’s audibility and effect on the perceived sound quality of the loudspeaker.
Isn't a sweep used for such purpose? This could be plotted as a continuous function without a limited number of intermediate steps.
Exactly. I have some speaker testing software for a mac -fuzzmeasure. You hook a mic up and run an audio output to the preamp and when you trigger it, you get what sounds like a swept wave of increasing frequency. The software records the speakers output via the mic and gives you a nice graph of your speakers frequency response.
Isn't a sweep used for such purpose? This could be plotted as a continuous function without a limited number of intermediate steps.
<snip>So the observed phenomenon is perfectly plausible without the need of a frequency response fraud conspiracy theory about a manufacturer you probably don't even know.
The day a majority of major manufactures publish response curves with 1/10th the resolution as those published at www.soundstageav.com is the day I stop believing they are sampling every half or even octave.
Oh, maybe there was a misunderstanding. I did not want to say that the frequency response was as flat as +/- 0.15db. That would be quite a value for car audio. I did want to say that the two plots did not deviate from each other by more than that amount.
On a related note:
If you compared the frequency response from your loudspeaker in a dead room in contrast to the frequency response when measured in your living room, I think you will find that you should start modifying your room before you spend a fortune on new speakers
Because a some people here seem to be such fans of Sean Olive I googled him. In his last article (http://seanolive.blogspot.com/2009/01/what-loudspeaker-specifications-are.html) about speaker quality metrics he's confirming what I conjectured in my upper post about distortion and a lack of usable metrics, regarding their relationship to a speaker's sound. So the observed phenomenon seems perfectly plausible without the need of a frequency response fraud conspiracy theory about a manufacturer you probably don't even know.
3. The relationship between perception and measurement of nonlinear distortions is less well understood and needs further research. Popular specifications like Total Harmonic Distortion (THD) and Intermodulation Distortion (IM) do not accurately reflect the distortion’s audibility and effect on the perceived sound quality of the loudspeaker.
This is the second time ina week I've seen someone quote *ONLY* point number three from Sean's blog, while either downplaying or ignoring the points that were #1 and #2:
#1 The perception of loudspeaker sound quality is dominated by linear distortions, which can be accurately quantified and predicted using a set of comprehensive anechoic frequency response measurements (see my previous posting here)
#2 Both trained and untrained listeners tend to prefer the most accurate loudspeakers when measured under controlled double-blind listening conditions (see this article here).
These points are *at least*, if not more, important than #3, fo someone seeking to buy loudspeakers on a rational basis . They indicate that we CAN make predictions about loudspeaker sound quality, from the right measurements (whihc, unfortunately, aren't always available). Point #3 adds that we don't have every correlated metric for every form of distortion worked out...yet.
In your post 'up there' you gave only the barest technical details; we don't knwo how the measurements were taken; we don't even know what the models were. The speakers were almost certainly not heard at the same position (even small changes in speaker position will change apparent treble performance); and of course the comparison wasn't blind. It's hard to say *anything* concrete about your anecdote, and implying that something mysterious and as yet unmeasureable was the reason for what you heard, is really jumping the gun..
If the only audible difference left between different digital systems is with the speakers,
Btw, the 'no audible difference' proviso applies when the 'digital system' output devices being compared are level matched (preferably to within 0.2dB) on all channels. This is by no means guaranteed to be the case 'out of box'.
No one should say 'all digital systems sound the same', without qualification.
This is the second time ina week I've seen someone quote *ONLY* point number three from Sean's blog, while either downplaying or ignoring the points that were #1 and #2:
Point #1 and #2 are just not news in any way. Condition one for a good speaker is a flat frequency response from x to at least 20000, with x as small as you can afford. That will already take you a very long way. But as said, that's obvious. The interesting point for debate is #3, because we are lacking appropriate metrics, yet.
This is the second time ina week I've seen someone quote *ONLY* point number three from Sean's blog, while either downplaying or ignoring the points that were #1 and #2:
Point #1 and #2 are just not news in any way. Condition one for a good speaker is a flat frequency response from x to at least 20000, with x as small as you can afford. That will already take you a very long way. But as said, that's obvious. The interesting point for debate is #3, because we are lacking appropriate metrics, yet.
That other guy that krabby is referring to is me, from a post I made over on the AVS forum. Funny that I would end up finding this thread too. I spend too much time on the interweb me thinks.
I too find point #3 the most interesting because I've been studying FR graphs for years, and have had plenty of speakers in my home with respectably flat FR, yet I would hear qualities to the sound of speakers that did not line up with their FR. Furthermore, I recently traded speakers with +/- 1dB FR from 50hz to 20KHz (Revel Studio2) for speakers with +/- 3dB FR (Wilson Benesch) because the later sounded more natural to me (and besides I can change the speaker's FR easily with DSP based EQ, but have no control over the distortion).
But the answer to the OPs question is simple. If you believe Olive and Toole's findings, that humans prefer speakers that are more accurate (flat on and off-axis FR, low distortion, etc), then you will naturally pick the more accurate speaker in a blind test simply by listening to them. No need to refer to graphs and measurements, because current testing shows that we will sort these things out with listening. OK, so you could narrow down the candidates by referring to the FR and dispersion graphs, but I wouldn't be too anal about it. +/- 3dB FR, given it is relatively smooth in its transitions should put the speaker in the running. Best to do the auditioning at home though, as the room, the listening position and speaker position can affect the outcome, as I'm sure you all know.
This is the second time ina week I've seen someone quote *ONLY* point number three from Sean's blog, while either downplaying or ignoring the points that were #1 and #2:
Point #1 and #2 are just not news in any way. Condition one for a good speaker is a flat frequency response from x to at least 20000, with x as small as you can afford.
What's 'news' depends on how clued in you are, doesn't it? Point #3 was not news to me....nor to anyone who followed the research that yielded #1 and #2 ...and most audio hobbyists would not be surprised to know that loudspeakers measurements aren't fully comprehensive predictors of sound yet.
(Flat FR on
and off axis matters...the latter is still news to some people. And FR measurments aren't the only measurements that are currently taken, that matter. Cabinet resonance,
for example)
But the answer to the OPs question is simple. If you believe Olive and Toole's findings, that humans prefer speakers that are more accurate (flat on and off-axis FR, low distortion, etc), then you will naturally pick the more accurate speaker in a blind test simply by listening to them. No need to refer to graphs and measurements, because current testing shows that we will sort these things out with listening. OK, so you could narrow down the candidates by referring to the FR and dispersion graphs, but I wouldn't be too anal about it. +/- 3dB FR, given it is relatively smooth in its transitions should put the speaker in the running. Best to do the auditioning at home though, as the room, the listening position and speaker position can affect the outcome, as I'm sure you all know.
Because few can do blind tests at home, for most there is certainly a need to refer to graphs and measurements , if they aim to own a loudspeaker that conforms to the Toole/Olive metrics for 'likely to sound good'
(dupe, delete)
(anotehr dupe, please delete)
Because few can do blind tests at home, for most there is certainly a need to refer to graphs and measurements , if they aim to own a loudspeaker that conforms to the Toole/Olive metrics for 'likely to sound good'
Here's something interesting. Not entirely conclusive, but it relevant to the topic of discussion.
Audioholics recently did a blind test "shoot out" between 4 speakers in the $1500-$2000 range, one of which was the new Infinity Classia C336. As Infinity is a brand under the Harman umbrella, the Classia was likely designed with flat frequency response as a goal, and they were probably subjected to internal blind testing.
Based on the FR graph provided by HT mag, it looks like Infinity hit the goal of flat FR. Well at least within a 30 degree window (HT Mag does not publish off-axis measurements unfortunately).
(http://hometheatermag.com/images/archivesart/409infsp.meas.jpg)
Despite the flat FR of the Classia, it was rated the least desirable in terms of sound quality. The group chose a Dali speaker consistently over the Infinity, and I have never seen a Dali speaker that has response anywhere near as flat is the Infinity.
http://www.audioholics.com/reviews/speaker...t-results-blind (http://www.audioholics.com/reviews/speakers/floorstanding/1500-2000-floor-shootout-2009/shootout-results-blind)
If you were to narrow down your choices using the FR plots with the speakers in this test, the Infinity's would make the short list, and the Dali's probably wouldn't.
Just food for thought.
A manufacturer that is striving for flat frequency response and perfect unit pulse response is Nubert www.nubert.de
Their philosophy is that a speaker cannot know, how to alter the sound as it cannot distinguish between a silent part in song at high volume or a loud part played at low volume. So the best one can do is just to reproduce faithfully what the source contains.
They also have lots of technical material to read on their page - unfortunately only in German.
Here's something interesting. Not entirely conclusive, but it relevant to the topic of discussion.
Audioholics recently did a blind test "shoot out" between 4 speakers in the $1500-$2000 range,...
hum...
There were two yes/no questions - Do you like this speaker? and Would you want to own this speaker? As you might imagine, the responses to both these questions lined up pretty closely. The thought was that you could like a speaker in a comparison but not want to own either of them. It would be very unusual to want to own a speaker that you didn't like. Results of those questions in the table below:
...
Each speaker was heard three times by each listener. The number is the absolute number of times the listener either liked or would buy the speaker. The maximum would be nine (all three saying yes in all three comparisons)
Sorry but that's really poor methodology, and is an entirely too small of a sample size to mean anything conclusively. Why wouldn't they try a standard ABC/HR (http://www.rarewares.org/rja/ListeningTest.pdf) test if they were going to waste their time?
I second the opinion about Nubert. It's an excellent manufacturer. They claim though, that their crossovers are of maximum fidelity while maintaining phase accuracy (through 'phase correction filters') which is theoretically impossible.
Sorry but that's really poor methodology, and is an entirely too small of a sample size to mean anything conclusively. Why wouldn't they try a standard ABC/HR (http://www.rarewares.org/rja/ListeningTest.pdf) test if they were going to waste their time?
I really don't think that something that rigorous is needed to blind test speakers. The differences are often quite obvious, as compared to testing something like a CD player or a preamp. Then something like the ABC test would be warranted.
@rpp3po: I think what you mean is only employed in the new Nubert nuVero series. And the filter math was shown in their forum. You may check the math whether it is impossible or not: http://www.nubert-forum.de/nuforum/viewtop...ro&start=10 (http://www.nubert-forum.de/nuforum/viewtopic.php?t=21027&postdays=0&postorder=asc&highlight=filter+nuvero&start=10)
Despite the flat FR of the Classia, it was rated the least desirable in terms of sound quality. The group chose a Dali speaker consistently over the Infinity, and I have never seen a Dali speaker that has response anywhere near as flat is the Infinity.
I'm a fan of what Sean Olive and researchers like him are doing, but I do wonder about something. A basic adage (and I believe there has been research in psychology to support this) is that people tend to prefer what they are used to. From my experience, this holds true for choices ranging from underwear type to the unconscious choices made when picking friends and romantic partners.
So, if a group of listeners have spent many years listening to speakers that are do NOT have flat FR, wouldn't they pick a similar sounding speaker as better sounding than a flat speaker in a blind test? Isn't it just a matter of liking what they are used to?
Honestly, I support most results of blind testing in audio, but it's very difficult for me to accept that Olive's results are representative of the larger population and not just his sample.
Good point.
Despite the flat FR of the Classia, it was rated the least desirable in terms of sound quality. The group chose a Dali speaker consistently over the Infinity, and I have never seen a Dali speaker that has response anywhere near as flat is the Infinity.
I'm a fan of what Sean Olive and researchers like him are doing, but I do wonder about something. A basic adage (and I believe there has been research in psychology to support this) is that people tend to prefer what they are used to. From my experience, this holds true for choices ranging from underwear type to the unconscious choices made when picking friends and romantic partners.
So, if a group of listeners have spent many years listening to speakers that are do NOT have flat FR, wouldn't they pick a similar sounding speaker as better sounding than a flat speaker in a blind test? Isn't it just a matter of liking what they are used to?
Honestly, I support most results of blind testing in audio, but it's very difficult for me to accept that Olive's results are representative of the larger population and not just his sample.
Did you read the papers to see what his sample consisted of? You might be surprised.
In addition to listeners from various occupations -- including audio 'professionals' -- Olive also compared listeners who he trained to hear typical forms of distortion, to those he did not. The former more strongly showed a preference for well-measuring speakers , than the latter -- but the trend was there in both groups.
These are trends, not perfect predictors of individual choice. Floyd Toole discusses some interesting departures from the main trend, in his book.
You could also write to Sean on his blog -- or his account here at HA, or on avsforum -- and ask him about variance due to previous experience. That would be an interesting discussion too
If you were to narrow down your choices using the FR plots with the speakers in this test, the Infinity's would make the short list, and the Dali's probably wouldn't.
Just food for thought.
First, it's results from just three listeners, whereas Olive tested hundredsn, and Audioholics, though they deserve all credit for trying, doesn't have the terrific 'double-blind speaker lazy susan' technology that Harman has, to reduce the interval between presentations and variation in speaker position. The room is also acoustically well-designed and treated. The Audioholics loudspeakers weren't auditioned from the same speaker placement, and the room by the reviewers' own account had some issues.
Second, given that two of the three subjects were audio reviewers, these results may well fall into Toole's secondary category of listeners who do NOT tend to prefer the sound of loudspeakers that meet the derived Toole/Olive criteria.
Those in this category tended to be audio engineers and others who make their living listening to audio in somewhat different ways than the typical listener. Instead of wide 'apparent source width' (ASW) and 'listener envelopment',
two key criteria for typical listeners, they tend to focus on things like pinpoint imaging.
Third, the subjects even seem to have discussed or viewed each other's answers during the test. That's a no-no.
Did you read the papers to see what his sample consisted of? You might be surprised.
In addition to listeners from various occupations -- including audio 'professionals' -- Olive also compared listeners who he trained to hear typical forms of distortion, to those he did not. The former more strongly showed a preference for well-measuring speakers , than the latter -- but the trend was there in both groups.
These are trends, not perfect predictors of individual choice. Floyd Toole discusses some interesting departures from the main trend, in his book.
I did go through one paper and it was precisely the variety of the listeners in his sample and the results that surprised me.
Honestly, It's not that I don't believe him, I just find it hard to digest. A lot of people (me included) were raised on systems with V shaped FRs. I remember when standalone"Graphic Equalizers" with fancy LCD displays were common with every home system and many people would immediately hit the button that boosted the bass and treble to "improve" the sound. Consequently, I tend to like headphones/speakers with a little bump in the bass/midbass.
If you were brought up listening to smaller portable boxes with one physical driver per channel, you probably got used to sound that is almost an inverse V. Wouldn't people raised that way prefer a speaker with a more humped up midrange?
I'm just speculating here. There's probably another theory that explains Olive's results and crushes my objections. I'm not the one doing research.
Did you read the papers to see what his sample consisted of? You might be surprised.
Honestly, It's not that I don't believe him, I just find it hard to digest. A lot of people (me included) were raised on systems with V shaped FRs. I remember when standalone"Graphic Equalizers" with fancy LCD displays were common with every home system and many people would immediately hit the button that boosted the bass and treble to "improve" the sound. Consequently, I tend to like headphones/speakers with a little bump in the bass/midbass.
What you probably don't know is the frequency response of the systems you are listening to *before* you crank in the bass boost.
af anyone wants to read an, um, interesting juxtaposition of subjective impression versus measurements of loudspeakers, check out this review:
http://www.positive-feedback.com/Issue11/zudruid.htm (http://www.positive-feedback.com/Issue11/zudruid.htm)
and then check out these measurements
http://www.soundstagemagazine.com/measurem.../zucable_druid/ (http://www.soundstagemagazine.com/measurements/zucable_druid/)
Btw, it's a loudspeaker by a company that also sells 'audiophile' cables.
Because few can do blind tests at home, for most there is certainly a need to refer to graphs and measurements , if they aim to own a loudspeaker that conforms to the Toole/Olive metrics for 'likely to sound good'
Here's something interesting. Not entirely conclusive, but it relevant to the topic of discussion.
Audioholics recently did a blind test "shoot out" between 4 speakers in the $1500-$2000 range, one of which was the new Infinity Classia C336. As Infinity is a brand under the Harman umbrella, the Classia was likely designed with flat frequency response as a goal, and they were probably subjected to internal blind testing.
Based on the FR graph provided by HT mag, it looks like Infinity hit the goal of flat FR. Well at least within a 30 degree window (HT Mag does not publish off-axis measurements unfortunately).
(http://hometheatermag.com/images/archivesart/409infsp.meas.jpg)
Despite the flat FR of the Classia, it was rated the least desirable in terms of sound quality. The group chose a Dali speaker consistently over the Infinity, and I have never seen a Dali speaker that has response anywhere near as flat is the Infinity.
http://www.audioholics.com/reviews/speaker...t-results-blind (http://www.audioholics.com/reviews/speakers/floorstanding/1500-2000-floor-shootout-2009/shootout-results-blind)
If you were to narrow down your choices using the FR plots with the speakers in this test, the Infinity's would make the short list, and the Dali's probably wouldn't.
Just food for thought.
af anyone wants to read an, um, interesting juxtaposition of subjective impression versus measurements of loudspeakers, check out this review:
http://www.positive-feedback.com/Issue11/zudruid.htm (http://www.positive-feedback.com/Issue11/zudruid.htm)
and then check out these measurements
http://www.soundstagemagazine.com/measurem.../zucable_druid/ (http://www.soundstagemagazine.com/measurements/zucable_druid/)
Btw, it's a loudspeaker by a company that also sells 'audiophile' cables.
KrabApple - This is probably one of the worst set of loudspeaker frequency response curves I've seen in quite awhile. When measurements look that bad, yet the listening review is positive, it's a sign that something is faulty with the listening device(s) and/or listening test methodology. It's a case of "Hearing is Believing" versus "Believing is Hearing." It wouldn't be the first time I've seen this.
The best example is a high-end $11k electrostatic speaker that was awarded "Speaker of the Year" in a well-known audio magazine, and yet it measured horribly. When I invited 6 professional audio reviewers (including some from the same magazine) to evaluate it under controlled, double-blind conditions they rated it dead last -- just like the other 300 listeners (see speaker M here (http://seanolive.blogspot.com/2008/12/part-3-relationship-between-loudspeaker.html) . The other speaker P that they rated in first place, was also deemed "speaker of the year" in the same magazine the following year. How can reviewers be so grossly inconsistent from year to year? It's because they don't do comprehensive measurements like SoundStage or now Consumer Report, but instead rely on poorly controlled, sighted, biased, casual listening. KrabApple's example is another good illustration of what can happen.
Cheers
Sean
http://seanolive.blogspot.com (http://seanolive.blogspot.com)
On a related note:
If you compared the frequency response from your loudspeaker in a dead room in contrast to the frequency response when measured in your living room, I think you will find that you should start modifying your room before you spend a fortune on new speakers
True for steady state.
Remember that the perception of the frequency response of a loudspeaker/room system has a complex interaction between the sound directly from the speaker to the listener followed by the arrival of the reverberation from the room/speaker interaction. So if the direct sound is perfectly flat, the time-delayed arrivals at the listener after the indirect sound bounces around the room and decays certainly won't be flat, so how the brain integrates these disparate signals is far more complex than a simple anechoic steady state frequency response could predict.
Another item that can have a significant impact is amplitude linearity. Most speakers tend to compress the amplitude as the sound level increases (some extremely so, and all speakers vary in this affect by frequency as well), and distortion can climb quickly as well. This is not often measured for home speakers (though it is for professional/monitors), but can have a significant impact on the sound of the speaker. If I remember correctly, many home speaker designs are actually pretty poor in this regard (can't remember where I read the test data...)
These are just some of the reasons there is no "best" objective speaker. That said, there are clearly more accurate designs, and room interaction is very important.
Did anybody notice that posts 48 and 49 are from THE Sean Olive we are talking about in this thread?
That being said, is there a reason there is no text in either post other than quotations from other posts? Did something go wrong when he was posting? I, for one, would like to know what Sean Olive wants to say!
Sorry, I didn't realize I had posted a blank. I have edited my posting for post 49, and will respond to 48 after I gather more information.
Cheers
Sean
Moderation: Removed unnecessary quotation of the previous post. Removed link to your site; what's in your signature should suffice. I removed post 48, but decided to put it back so that you may edit it in place.
Despite the flat FR of the Classia, it was rated the least desirable in terms of sound quality. The group chose a Dali speaker consistently over the Infinity, and I have never seen a Dali speaker that has response anywhere near as flat is the Infinity.
I'm a fan of what Sean Olive and researchers like him are doing, but I do wonder about something. A basic adage (and I believe there has been research in psychology to support this) is that people tend to prefer what they are used to. From my experience, this holds true for choices ranging from underwear type to the unconscious choices made when picking friends and romantic partners.
So, if a group of listeners have spent many years listening to speakers that are do NOT have flat FR, wouldn't they pick a similar sounding speaker as better sounding than a flat speaker in a blind test? Isn't it just a matter of liking what they are used to?
Honestly, I support most results of blind testing in audio, but it's very difficult for me to accept that Olive's results are representative of the larger population and not just his sample.
It think it is generally true that people tend to prefer what they are used to, yet that has not been my experience with loudspeaker tests. There was a 1956 study published in JASA by Kirk (http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JASMAN000028000006001113000001&idtype=cvips&gifs=yes) who reported that college students preferred band-limited loudspeakers over a wider bandwidth loudspeaker. The students only preferred the better loudspeaker after they had been exposed to it over a period of time. There are some problems with that study, one of them being that the poor quality of the source material from that era may have been a biasing factor: the hiss and distortions may have actually sounded better on band-limited loudspeakers. Still, the results cannot be entirely discounted.
Fast forward 53 years later (2009) and a similar sort of study is being reported. A Stanford music professor (http://www.audioholics.com/news/industry-news/kids-prefer-poor-quality-mp3?date=100220090311) reports that his college students prefer low bit-rate MP3 music over the higher quality lossless CD quality version; he believes it is because that is what the students are used to hearing. If you consider that most of the headphones and speakers sold with MP3 players today are not very good either, the future of high quality audio seems rather bleak
In 2003, I published a loudspeaker preference study (http://www.aes.org/tmpFiles/elib/20090406/12206.pdf) which included a small sample of college students along with trained listeners, audio reviewers, audio retailers, and audio marketing and sales people. The college students were the least discriminating and reliable (http://seanolive.blogspot.com/2008/12/part-2-differences-in-performances-of.html) group of listeners in the study. This could be explained alone by their young age and limited experience and exposure to good audio. Fortunately, they still preferred the more accurate loudspeakers over the least accurate one, but they could not reliably choose between the top three, which they tended to give very high ratings compared to the other groups of listeners. I'm sure many of these listeners in the study had bad loudspeakers at home, yet they chose the more accurate speakers in this study.
Finally, I use myself as a good example of someone who didn't prefer what I am used to. When I first started working at NRC back in 1985, someone slipped my own speakers into a loudspeaker test without my knowledge. These were speakers that I had owned and listened to for hundreds of hours over the past 7 years in college. After completing the listening test, I found out to my horror that I rated my own speakers dead last, finding lots of faults with them compared to the more accurate loudspeakers. It seems that within the context of a properly controlled multiple comparison loudspeaker test, people have an innate ability to separate the good loudspeakers from the poor ones.. and they tend to prefer the ones with the least offensive distortions.
If you were to narrow down your choices using the FR plots with the speakers in this test, the Infinity's would make the short list, and the Dali's probably wouldn't.
Just food for thought.
First, it's results from just three listeners, whereas Olive tested hundredsn, and Audioholics, though they deserve all credit for trying, doesn't have the terrific 'double-blind speaker lazy susan' technology that Harman has, to reduce the interval between presentations and variation in speaker position. The room is also acoustically well-designed and treated. The Audioholics loudspeakers weren't auditioned from the same speaker placement, and the room by the reviewers' own account had some issues.
Second, given that two of the three subjects were audio reviewers, these results may well fall into Toole's secondary category of listeners who do NOT tend to prefer the sound of loudspeakers that meet the derived Toole/Olive criteria.
Those in this category tended to be audio engineers and others who make their living listening to audio in somewhat different ways than the typical listener. Instead of wide 'apparent source width' (ASW) and 'listener envelopment',
two key criteria for typical listeners, they tend to focus on things like pinpoint imaging.
Third, the subjects even seem to have discussed or viewed each other's answers during the test. That's a no-no.
We have never said that audio reviewers "do NOT tend to prefer the the sound of loudspeakers that meet the derived Toole/Olive criteria." In fact, our data suggests that they tend to prefer the same loudspeakers as our trained listeners when given the opportunity to review them under well-controlled double-blind conditions (click on the graph here (http://seanolive.blogspot.com/2008/12/loudspeaker-preferences-of-trained.html) and compare the audio reviewers' preferences to the other groups' ). The problem is that audio reviewers seldom evaluate loudspeakers under properly controlled double-blind conditions, which means we rarely read a valid, unbiased opinion about the sound quality of the product.
Based on that same study, we did find that the audio reviewers as a group were about 1/5 as discriminating (http://seanolive.blogspot.com/2008/12/part-2-differences-in-performances-of.html) and reliable as the trained listeners. Age and occupational-related hearing loss could have been a factor since this sample of reviewers were on average older than the other groups of listeners. This is pure speculation since we only tested the hearing of the trained listeners, who had normal hearing.
Toole on page 436 in his new book " Sound Reproduction" talks about hearing loss, which are unfortunate afflictions, "especially for audio professionals and audio reviewers whose judgments are no longer representative of normal hearing listeners." He concludes: "an audiogram should perhaps be part of the personal resume of people in certain sensitive areas of the audio business, displaying evidence of why anyone should trust their opinions about sound quality."
Sean,
A couple of things spring to mind.
How does the average age of the audio reviewer compare to the average age of the reader of audio reviews? It could be that their profiled hearing perfectly matches the hearing of those who buy loudspeakers on the basis of reviews in audio magazines. The last hi-fi show I attended, Methuselah would have brought the average age of attendees down by about 30 years.
I'm also wondering how much of the audio reviewers poor score comes down to cognitive dissonance. If a reviewer holds the belief that blind tests are valueless (or worse, harmful to their career), engaging in a blind test will be a more uncomfortable experience for them than for other test subjects. People like Robert Harley seem to be very 'anti' any kind of blind testing and that might carry over into the test itself. While I don't think anyone would actively try to skew the test (even if they could), perhaps the audio reviewers are trying too hard. They have the most to lose by 'failing'; their own reputation as golden-eared reviewer is at stake, even if only to themselves. This could possibly undermine their performance. I have absolutely no idea how you'd test for that, however.
Personally, I think it might be that the audio reviewer group spend too long listening to the audio marketing and sales group and not enough time listening to speakers.
Moderation: Removed unnecessary quotation of the previous post.
af anyone wants to read an, um, interesting juxtaposition of subjective impression versus measurements of loudspeakers, check out this review:
http://www.positive-feedback.com/Issue11/zudruid.htm (http://www.positive-feedback.com/Issue11/zudruid.htm)
and then check out these measurements
http://www.soundstagemagazine.com/measurem.../zucable_druid/ (http://www.soundstagemagazine.com/measurements/zucable_druid/)
Btw, it's a loudspeaker by a company that also sells 'audiophile' cables.
KrabApple - This is probably one of the worst set of loudspeaker frequency response curves I've seen in quite awhile. When measurements look that bad, yet the listening review is positive, it's a sign that something is faulty with the listening device(s) and/or listening test methodology. It's a case of "Hearing is Believing" versus "Believing is Hearing." It wouldn't be the first time I've seen this.
Cheers
Sean
http://seanolive.blogspot.com (http://seanolive.blogspot.com)
The problem I have here is that this speaker may get a pasting in the lab, but gets lots of light and fluffy hyperbole in the review ON THE SAME WEBSITE!
How can this:
http://www.soundstagemagazine.com/measurem.../zucable_druid/ (http://www.soundstagemagazine.com/measurements/zucable_druid/)
Be this:
http://www.soundstage.com/revequip/zucable_druid.htm (http://www.soundstage.com/revequip/zucable_druid.htm) ?
I suppose at least the reviewer admitted that 'bass power was nearly absent'.
We have never said that audio reviewers "do NOT tend to prefer the the sound of loudspeakers that meet the derived Toole/Olive criteria." In fact, our data suggests that they tend to prefer the same loudspeakers as our trained listeners when given the opportunity to review them under well-controlled double-blind conditions (click on the graph here (http://seanolive.blogspot.com/2008/12/loudspeaker-preferences-of-trained.html) and compare the audio reviewers' preferences to the other groups' ). The problem is that audio reviewers seldom evaluate loudspeakers under properly controlled double-blind conditions, which means we rarely read a valid, unbiased opinion about the sound quality of the product.
Based on that same study, we did find that the audio reviewers as a group were about 1/5 as discriminating (http://seanolive.blogspot.com/2008/12/part-2-differences-in-performances-of.html) and reliable as the trained listeners. Age and occupational-related hearing loss could have been a factor since this sample of reviewers were on average older than the other groups of listeners. This is pure speculation since we only tested the hearing of the trained listeners, who had normal hearing.
Toole on page 436 in his new book " Sound Reproduction" talks about hearing loss, which are unfortunate afflictions, "especially for audio professionals and audio reviewers whose judgments are no longer representative of normal hearing listeners." He concludes: "an audiogram should perhaps be part of the personal resume of people in certain sensitive areas of the audio business, displaying evidence of why anyone should trust their opinions about sound quality."
Sean,
A couple of things spring to mind.
How does the average age of the audio reviewer compare to the average age of the reader of audio reviews? It could be that their profiled hearing perfectly matches the hearing of those who buy loudspeakers on the basis of reviews in audio magazines. The last hi-fi show I attended, Methuselah would have brought the average age of attendees down by about 30 years.
I'm also wondering how much of the audio reviewers poor score comes down to cognitive dissonance. If a reviewer holds the belief that blind tests are valueless (or worse, harmful to their career), engaging in a blind test will be a more uncomfortable experience for them than for other test subjects. People like Robert Harley seem to be very 'anti' any kind of blind testing and that might carry over into the test itself. While I don't think anyone would actively try to skew the test (even if they could), perhaps the audio reviewers are trying too hard. They have the most to lose by 'failing'; their own reputation as golden-eared reviewer is at stake, even if only to themselves. This could possibly undermine their performance. I have absolutely no idea how you'd test for that, however.
Personally, I think it might be that the audio reviewer group spend too long listening to the audio marketing and sales group and not enough time listening to speakers.
Gag,
According to Stereophile (http://www.stereophile.com/asweseeit/106awsi/) the median readership age was 48 years in 2006, and increasing annually. Your point is well taken: the hi-fi reviewers' median age may accurately reflect their readership. However,hearing loss is a very nonlinear process and I doubt you can categorize loudspeaker preferences of people with hearing loss into a single segment (unless the loudspeaker is a hearing aid ). In fact, Toole's data indicate that listeners with hearing loss as a group tend to have poor inter and intra-listener reliability in their ratings (their preferences are all over the map from one day to the next)
Your second point is that some reviewers and self-proclaimed golden ears may under perform in controlled listening tests because of "stage fright", fear of failure, or it may challenge their faith-based audio beliefs. Our trained listeners at Harman get over this psychological barrier through experience in tests, and participating in our listener training program, which teaches them how to rate sound quality attributes with constant feedback given on their performance. If after several weeks the listeners cannot make the grade, we reject them. Life is cruel. But if you want to be a doctor, airline pilot, concert pianist, or an Olympic athlete, you have to pass exams and win qualification races in order to demonstrate you "have the right stuff." Why should being a Harman listener or a professional audio reviewer be any different?
Sean,
Thanks for getting back to me so quickly.
Yes, age is a blunt tool to determine a non-linear process like hearing loss, but there are broad generalities that can be drawn from an aging population. This also makes me wonder; not every loudspeaker company uses a thorough set of measurement protocols to develop loudspeakers - some of the smaller companies designs come down to the 'golden-eared' skills of an individual. As these individuals age, do their loudspeaker designs come to rely more on measurement, or their failing hearing?
On your second point, in fairness to the audio reviewers, did they get the opportunity to get over their own 'stage fright' in tests? They were listed among the untrained, after all. It seems likely that once they got past that initial hurdle, they'd respond like any prospective trained band of listeners, with some failing to make the grade and some passing muster (age-related hearing loss and job-related hearing damage notwithstanding). Given the poor showing of the reviewers, it would be interesting to see how they responded to the training program.
That being said, the 'age-related hearing loss and job-related hearing damage' part could be the source of the poor results from the audio reviewers, after all. This would likely de-select some of them from the training process.
Edit: I just keep thinking up extra stuff... sorry. I was wondering how - if - training in other fields influences the performance of the panelists? For example, a musician's ear training, or a studio engineer trained in critical listening.
Moderation: Removed unnecessary quote of the previous post. Please refrain from doing this in the future.
Sean,
Thanks for getting back to me so quickly.
Yes, age is a blunt tool to determine a non-linear process like hearing loss, but there are broad generalities that can be drawn from an aging population. This also makes me wonder; not every loudspeaker company uses a thorough set of measurement protocols to develop loudspeakers - some of the smaller companies designs come down to the 'golden-eared' skills of an individual. As these individuals age, do their loudspeaker designs come to rely more on measurement, or their failing hearing?
On your second point, in fairness to the audio reviewers, did they get the opportunity to get over their own 'stage fright' in tests? They were listed among the untrained, after all. It seems likely that once they got past that initial hurdle, they'd respond like any prospective trained band of listeners, with some failing to make the grade and some passing muster (age-related hearing loss and job-related hearing damage notwithstanding). Given the poor showing of the reviewers, it would be interesting to see how they responded to the training program.
That being said, the 'age-related hearing loss and job-related hearing damage' part could be the source of the poor results from the audio reviewers, after all. This would likely de-select some of them from the training process.
Edit: I just keep thinking up extra stuff... sorry. I was wondering how - if - training in other fields influences the performance of the panelists? For example, a musician's ear training, or a studio engineer trained in critical listening.
Moderation: Removed unnecessary quote of the previous post. Please refrain from doing this in the future.
I can't answer for other audio manufacturers, but normal hearing is a pre-requisite for all listeners on our listening panel. Products must pass a minimum performance standard based on objective measurements before they are even submitted for subjective evaluation. Many of the reviewers have visited Harman listening labs several times before, and have participated in similar tests before. I believe that audio reviewers would benefit from listener training, as most listeners do. In fact, the founder of Stereophile magazine, Gordon Holt said in a recent interview (http://www.stereophile.com/asweseeit/1107awsi/) that the high-end audio establishment's failure to adopt blind product testing (ear training being part of that) has, in part, been responsible its lack of credibility and slow demise. I quote him:
"...Audio as a hobby is dying, largely by its own hand. As far as the real world is concerned, high-end audio lost its credibility during the 1980s, when it flatly refused to submit to the kind of basic honesty controls (double-blind testing, for example) that had legitimized every other serious scientific endeavor since Pascal. [This refusal] is a source of endless derisive amusement among rational people and of perpetual embarrassment for me, because I am associated by so many people with the mess my disciples made of spreading my gospel "
".. Remember those loudspeaker shoot-outs we used to have during our annual writer gatherings in Santa Fe? The frequent occasions when various reviewers would repeatedly choose the same loudspeaker as their favorite (or least-favorite) model? That was all the proof needed that [blind] testing does work, aside from the fact that it's (still) the only honest kind. It also suggested that simple ear training, with DBT confirmation, could have built the kind of listening confidence among talented reviewers that might have made a world of difference in the outcome of high-end audio. "
Cheers
Sean
My Audio Blog (http://seanolive.blogspot.com)
I'd like to know what's objectively the best speaker.
Simply put ... the best speaker is what sounds best to your ears with music of your choice.
There does exist a large amount of "construction philosophies" when it comes to speaker design. Some people enjoy smaller "close-field monitor speakers" whereas other listeners need a solid bass reproduction that normally can only be achieved with larger "floorstanders" together with proper cabinet ventilation or subwoofer/satellite systems.
Some people even go to extremes and choose rather exotic speaker designs like full-range horn speakers (have you ever seen a real bass horn?) or electrostats/magnetostats or classic desings with high-tech stuff like plasma tweeters or tweeters with diamond membranes.
Basically, choosing a loudspeaker goes along with answering questions like:
- What's my overall budget?
- How large is the room to put them in or how large can the speakers be?
- Where will I put my speakers?
- What kind of speaker design do I like?
- What kind of music do i mostly listen to? What intruments are my favourite ones?
- What kind of wattage does my amp deliver or what kind of speaker efficiency is needed for sufficient listening volume?
- What kind of listening volumes must be achieved?
- Continued ...
There is no such thing as a general-purpose loudspeaker that will excel at every aspect. Each desing principle (heck, even each membrane material) has it's advantages and shortcomings. As a result, buying speakers does always mean dealing with some sort of compromise.
As a result, buying speakers does always mean dealing with some sort of compromise.
Is that what they're calling salesmen these days?
I'd like to know what's objectively the best speaker.
Simply put ... the best speaker is what sounds best to your ears with music of your choice.
My question wasn't what would be the best speaker for me or my room, but rather, whether there was a speaker, or speaker design, that was the most able to accurately reproduce the original sound in its given environment.
Fast forward 53 years later (2009) and a similar sort of study is being reported. A Stanford music professor (http://www.audioholics.com/news/industry-news/kids-prefer-poor-quality-mp3?date=100220090311) reports that his college students prefer low bit-rate MP3 music over the higher quality lossless CD quality version; he believes it is because that is what the students are used to hearing. If you consider that most of the headphones and speakers sold with MP3 players today are not very good either, the future of high quality audio seems rather bleak
I wouldn't worry too much based on that news report. In a half-dozen or so different versions of it, I've yet to see any details on that music professor's 'study' that indicates it was worth a damn...like whether the comparisons were blind.
(Though of course that hasn't stopped the online audio forums from going bonkers over it...)
Finally, I use myself as a good example of someone who didn't prefer what I am used to. When I first started working at NRC back in 1985, someone slipped my own speakers into a loudspeaker test without my knowledge. These were speakers that I had owned and listened to for hundreds of hours over the past 7 years in college. After completing the listening test, I found out to my horror that I rated my own speakers dead last, finding lots of faults with them compared to the more accurate loudspeakers. It seems that within the context of a properly controlled multiple comparison loudspeaker test, people have an innate ability to separate the good loudspeakers from the poor ones.. and they tend to prefer the ones with the least offensive distortions.
Did they sound as good to you after this illuminating trial, as they did before it?
Second, given that two of the three subjects were audio reviewers, these results may well fall into Toole's secondary category of listeners who do NOT tend to prefer the sound of loudspeakers that meet the derived Toole/Olive criteria.
Those in this category tended to be audio engineers and others who make their living listening to audio in somewhat different ways than the typical listener. Instead of wide 'apparent source width' (ASW) and 'listener envelopment',
two key criteria for typical listeners, they tend to focus on things like pinpoint imaging.
We have never said that audio reviewers "do NOT tend to prefer the the sound of loudspeakers that meet the derived Toole/Olive criteria." In fact, our data suggests that they tend to prefer the same loudspeakers as our trained listeners when given the opportunity to review them under well-controlled double-blind conditions (click on the graph here (http://seanolive.blogspot.com/2008/12/loudspeaker-preferences-of-trained.html) and compare the audio reviewers' preferences to the other groups' ). The problem is that audio reviewers seldom evaluate loudspeakers under properly controlled double-blind conditions, which means we rarely read a valid, unbiased opinion about the sound quality of the product.
It's my mistake for misremembering something from Floyd Toole's book. I was recalling the section(s) on 'sensitive listeners' (e.g., p 119) -- also referred to as 'audio professionals' -- I mistakenly thought I'd read that audio reviewers were included in that group, but looking at it again, it appears to refer only to recording and mixing engineers and musicians and acousticians.
It's also not referring to *loudspeaker* preference per se, but to preference for reduced lateral reflections, compared to the levels preferred by most of the tested listeners.
My question wasn't what would be the best speaker for me or my room, but rather, whether there was a speaker, or speaker design, that was the most able to accurately reproduce the original sound in its given environment.
There do exist speakers that combine what is believed to be essential towards accurate sound reproduction.
These characteristics include:
-a flat frequency response, even under given angles (see according graph)
-low chassis resonance from its coil (the impedance graph will hint at that) or membrane (other graphs can tell you about partial oscillation of the membrane in the frequency range)
-sophisticated frequency crossover network layouts with low-tolerance components that e.g. introduce less phase-shifting and allow to deal with chassis resonance by absorption circuits
-fast decay times over the whole frequency range (the waterfall graph is useful for that)
-step response of speaker chassis
-rigid and resonance-free cabinet layout (you don't want to actually hear the cabinet walls move)
If I were to design a speaker I would make use of both classic acoustic teachings as well as modern electronic technology.
My system would most likely turn out to be a three-way floorstander with dual 8" bass chassis in sealed compartments and a D'Appolito (or coax) midrange/high system.
I would build a rigid cabinet and choose chassis that combine low weight membranes, decent internal dampening membrane material, unproblematic impedance graphs and maximum membrane stiffness (to avoid partial oscillations). The loudspeaker would be a fully active (one amp for each way) system with digital (maybe HDMI) inputs (future-proof for potentially useless, but nonetheless upcoming 192/24 formats) and complete internal digital signal processing (basically consists of crossover network, bass eq, parametric eq and time alignment) up to the analog amp outputs.
The digital layout would allow for sophisticated signal processing, quite similar to techniques used in car stereo and home cinema systems and could also be used to get rid of influences by material tolerance.
Time alignment would be used to improve step response by making sure that each speaker chassis moves at the right time (under consideration of the actual listening position). A digital crossover network would allow to implement steep filtering without the disadvantages of passive crossover networks. A parametric eq (with sufficient bands for each speaker way) would allow to get rid of acoustic problems that might arise in the actual listening environment (like standing waves or resonating objects). Additionally, the parametric eq can be used to fine-tune the speaker's sound to personal preferences. The extra bass eq would be used to extend (or reverse-eq) the lower frequency range (sealed cabinets of lower volumes normally produce an early drop-off in frequency response but offer a good impulse response over ventilated systems).
As it is common with home-cinema systems, there would also be an auto-calibration, much similar to Alpine's car audio implementation called "Imprint".
By using a USB interface and supplying an easy-to-use control software, each customer would be able to control the speakers' sound via home computer. A useful addition could be analog inputs with ADC's ...
Just a few thoughts, though
Edit: a few additions
I wouldn't worry too much based on that news report. In a half-dozen or so different versions of it, I've yet to see any details on that music professor's 'study' that indicates it was worth a damn...like whether the comparisons were blind.
(Though of course that hasn't stopped the online audio forums from going bonkers over it...)
Did they sound as good to you after this illuminating trial, as they did before it?
1. See I would love to see the data from the MP3 study but the author doesn't respond to email. I can't believe that the media would go bonkers over the findings without referencing or reading a scientific paper about it.
2. I gave my old speakers (Genesis) away shortly after that illuminating test as I became exposed to a lot of good speakers while working at the NRC during those years (1985-93). At various times I worked for and tested speakers for companies like Paradigm, PSB and Energy/Mirage --
Cheers
Sean
Audio Musings (http://seanolive.blogspot.com)
2. I gave my old speakers (Genesis) away shortly after that illuminating test as I became exposed to a lot of good speakers while working at the NRC during those years (1985-93). At various times I worked for and tested speakers for companies like Paradigm, PSB and Energy/Mirage --
Would you feel inclined to mention a few of the brands that best matched the results (flat FR, good off-axis performance, etc) of your research?
I think I would like to add a few thoughts as well. I used to test speakers at Shure Inc. I was also a phono cartridge designer (shows my age).
One big problem with judging accuracy is that most music is manufactured. Nothing to compare it to!
Heck, even in my little studio I use eight microphones on one drum set. It's not even close to any kind of accurate sonic capture.
That being said, I would go for something that can produce sound over a wide frequency range with reasonably good distortion figures. I would not consider a good on axis anechoic response as proof of sonic accuracy, but a very bad curve might indicate problems.
One of the most important factors to me is output level capabilities of the system. I don't mean just to listen to very loud music. I'll give an example. A handclap at arms length can easily have a peak SPL of 120 db or more unweighted. It doen't really sound loud. But many if not most systems can't do that...a simple hand clap.
For me having one that can really affects the level of realism on the right recording, particularly with the 120 db+ dynamic range capabilities of some modern gear. Granted...we're talking about 24 bit studio stuff here rather than a cd player, but that kind of performance will be in consumer stuff too, if not already.
One other thing I wanted to mention. The BRS (binaural room scanning) written about in Sean's blog.
Long ago we did some testing with head movements in binaural recordings. It was not as sophisticated as binaural impulse response processing. It was simply having the listener's head move in the same way that a binaural dummy head did in the original sound field. It was uncanny. We simply could not tell what was recorded and what was live headphone bleed. If you walked down the hall and shouted "Hey, take off the headphones. Lets go eat lunch" during the recording the test subject would promply take off the phones and say ok, I'm coming, then head for the hall. Even though it was a recording!!!
That was the most accurate sound reproduction I have ever heard. Here's the rub...the astounding realism (not accuracy) still happened if the frequency response was severly limited and the phase response was strongly degraded. It even worked with telephone like quality.
That showed me just how far removed some speakers in a room are from any kind of sonic reality.
It's the format really. So I just use speakers that sound good and make me smile. They are neither flat nor "accurate", but they can reproduce low distortion audio over the audible range at very high SPL. To me this gives an effortless quality at normal low listening levels.
Les Watts
L M Watts Technology
2. I gave my old speakers (Genesis) away shortly after that illuminating test as I became exposed to a lot of good speakers while working at the NRC during those years (1985-93). At various times I worked for and tested speakers for companies like Paradigm, PSB and Energy/Mirage --
Would you feel inclined to mention a few of the brands that best matched the results (flat FR, good off-axis performance, etc) of your research?
This is ancient history (> 16 years ago), but most of the companies that used the NRC were aiming for flat FR & good off-axis performance because that is what the NRC research showed produced the highest listener fidelity ratings. The companies included the ones I mentioned -- Paradigm, PSB, Energy/Mirage -- and Camber, Axiom,Waveform Research - and few others. It was at NRC, where I met Kevin Voecks (now at Revel) who worked for Mirage and then Snell at that time. He would travel from Los Angeles to Ottawa just to measure and listen to his prototype speakers.
Cheers
Sean
Audio Musings (http://seanolive.blogspot.com)
One big problem with judging accuracy is that most music is manufactured. Nothing to compare it to!
One of the most important factors to me is output level capabilities of the system
One other thing I wanted to mention. The BRS (binaural room scanning) written about in Sean's blog.
Long ago we did some testing with head movements in binaural recordings. It was not as sophisticated as binaural impulse response processing. It was simply having the listener's head move in the same way that a binaural dummy head did in the original sound field. It was uncanny. We simply could not tell what was recorded and what was live headphone bleed.
Hi Les,
1. We try to pick timbrally accurate recordings that cover the entire audio bandwidth. Even though the recordings are mostly "manufactured" listeners become familiar with them and know how they should sound. In multiple A/B/C/D comparison tests, any errors in the recordings are common to all loudspeakers and tend to be ignored. By switching between different speakers you can quickly detect the presence of resonances and which speakers are not spectrally balanced.
2. I agree that output level capabilities are important, and the most difficult to test subjectively without damaging the listeners' hearing. I think objective tests are probably the best way to characterize power compression,etc. There is still no standard loudspeaker specification for nonlinear distortion that is perceptually meaningful, unfortunately.
3. We did a similar thing back at NRC using two chairs that could rotate via a stepper motor and were coupled to each other When the listener in another room rotated in their chair, the KEMAR sitting in the listening room in an identical chair would rotate the same angle. The realism was quite good but never as good as when the listener sat in the same room as KEMAR. It's important with auditory displays to have realistic visual cues so there is no cognitive dissonance between the senses. For that reason we do BRS listening tests in the same room or car (or a facsimile of the car) as where the scans were made. It adds to the realism/presence in the auditory display, particularly in terms of spatialization.
Cheers
Sean
Audio Musings by Sean Olive (http://seanolive.blogspot.com)
1. See I would love to see the data from the MP3 study but the author doesn't respond to email. I can't believe that the media would go bonkers over the findings without referencing or reading a scientific paper about it.
@solive--I think you have just increased your credibility as a dedicated researcher (as though that were needed) by showing that you waste no time at all on the media.
this sounds interesting:
http://techon.nikkeibp.co.jp/english/NEWS_...0090317/167297/ (http://techon.nikkeibp.co.jp/english/NEWS_EN/20090317/167297/)
sound-cancelling technology being used to make loudspeaker output more accurate
@rpp3po: I think what you mean is only employed in the new Nubert nuVero series. And the filter math was shown in their forum. You may check the math whether it is impossible or not: http://www.nubert-forum.de/nuforum/viewtop...ro&start=10 (http://www.nubert-forum.de/nuforum/viewtopic.php?t=21027&postdays=0&postorder=asc&highlight=filter+nuvero&start=10)
Nubert has been using linkwitz-riley crossover filters for their speakers for quite a while.
Their new filter solves problems inherent in designing crossovers for Satellite/Subwoofer systems and 3-Way Speakers (the sum of outputs have near perfect frequency and phase linearity), previously only possible with FIR DSPs (but without the long group delays associated with FIR filters with low frequencies). But other then what they thought it seems that the concept is not entirely new, so they might not get that patent...
If the only audible difference left between different digital systems is with the speakers, personally, I'd like to know what's objectively the best speaker.
To me, thinking about "the best speaker" is strange because IMO they are all so bad that comparing speakers is like judging mud pies at a county fair as if they were apple pies.
IOW, show me a speaker that is as good of a transducer operating in a room as an apple pie at a county fair, and then we can talk about which one is best. ;-)
-------
Moving down this thread, Les who used work at Shure comes closest to agreeing with my opinion of speakers in general. To me, some of them are good enough to work with given that they are all we have, but all the best speakers do is make pleasing noises. True realism or even a good approximation of it is nowhere to be heard.
Also near the end of the thread you see comments that stunning realism is possible with headphones and binaural recordings. I agree. One problem with binaural recordings is that they sound strange when played on speakers.
The idea of masses of people walking around listening to music with headphones used to be pretty strange, but now people do it all the time. Maybe we ought to just forget about using loudspeakers for most listening applications and just use headphones or earphones. It isn't hard to mix multichannel recordings so that they sound great on headphones. I sense that with the iPod revolution in full swing, that may become a trend.
IME there is one loudspeaker-based listening experience that is more like headphones, and that is listening in a car.
Of course there are situations where speakers seem to be more convenient. However, I notice some trends and that is more and more audiophiles seem to be iterating their listening rooms towards more and more acoustic treatments.
If this trend continues we might find ourself facing a new truth, and that the best speaker might be the one that least engages the listening room.
So, objectively speaking, what's the most accurate, or honest, speaker made?
Summa speakers, designed, built and sold by my good friend Earl Geddes.
http://www.gedlee.com/Summa.htm (http://www.gedlee.com/Summa.htm)
Only, when Earl's not looking, you need to beef up their bottom ends with subwoofers. ;-)
Which type of speaker design is the best overall?
As close to a full-range waveguide as you can get. In the real world we keep having to blend waveguides and direct radiators, but the transition to the DR should be done at the lowest frequency possible.
Is it even possible to build a 'transparent' speaker?
We do remarkably well with headphones and earphones.
How can you properly compare two sets of speakers?
Talk to Sean Olive.
He seems to be posting here, time to ask him your burning questions - but maybe read up on all of his online stuff first, before boring him with questions he has already answered. ;-)
By the way, personally, I'd vote for electrostatic speakers.
Electrostatic is just a transducer design, not a speaker system design.
You really didn't answer your own question. ;-)
So, objectively speaking, what's the most accurate, or honest, speaker made?
Summa speakers, designed, built and sold by my good friend Earl Geddes.
http://www.gedlee.com/Summa.htm (http://www.gedlee.com/Summa.htm)
Only, when Earl's not looking, you need to beef up their bottom ends with subwoofers. ;-)
AFAIK, Earl recommends the use of subs with all of his speakers, including the Summas.
So, objectively speaking, what's the most accurate, or honest, speaker made?
Summa speakers, designed, built and sold by my good friend Earl Geddes.
http://www.gedlee.com/Summa.htm (http://www.gedlee.com/Summa.htm)
Only, when Earl's not looking, you need to beef up their bottom ends with subwoofers. ;-)
AFAIK, Earl recommends the use of subs with all of his speakers, including the Summas.
I'm sure that's true of the smaller Summas, but Earl has 3 of his largest Summas in his listening room, and he has claimed in the past that said room has 3 subs in it due to their presence.
The woofer he uses in his big Summas is probably good down to 32 Hz, but 16 not so much.
So, objectively speaking, what's the most accurate, or honest, speaker made?
Summa speakers, designed, built and sold by my good friend Earl Geddes.
http://www.gedlee.com/Summa.htm (http://www.gedlee.com/Summa.htm)
Only, when Earl's not looking, you need to beef up their bottom ends with subwoofers. ;-)
AFAIK, Earl recommends the use of subs with all of his speakers, including the Summas.
I'm sure that's true of the smaller Summas, but Earl has 3 of his largest Summas in his listening room, and he has claimed in the past that said room has 3 subs in it due to their presence.
The woofer he uses in his big Summas is probably good down to 32 Hz, but 16 not so much.
I thought he has 3 Summas and 3 subs in his listening room.
The comments here would seem to back this up:
http://www.gedlee.com/downloads/recommendations.pdf (http://www.gedlee.com/downloads/recommendations.pdf)