We can do all sorts of in room or outside measurements to accurately define a speaker. Correct?

Then why do people still ignore all those measurements and use "How does it sound to you" Mantra.

Does accurate speaker equate to best measured speaker? (Flat FR plot, etc)

If the goal is accuracy for some people then why don't they use the measurements instead of the flawed "How does it sound to you" appoarch?

How it sounds to them has little to do with accuracy most times. I have to believe people just think there is a missing measurement out there and therefore they have to listen.

Obviously, Im assuming accurate measurements....yes manufacturers can lie!! Lets assume a speaker has the measurements you want. Why not buy it based on those?

How do you measure: Spaciousness? Size of Soundstage? Warmth? Bloated Sound? Muddiness? Detail? Openess? Boxiness? Accuracy is just a small part of the music equation.

How do you measure: Spaciousness? Size of Soundstage? Warmth? Bloated Sound? Muddiness? Detail? Openess? Boxiness? Accuracy is just a small part of the music equation.

You are saying those are not in measurements.

Off axis is sound stage/spaciousness/boxiness/openess/spaciousness/soundstage

Decay plot is reverberation/bloated sound/muddiness

Warmth is a slop downward on the FR plot.

Detailed is a flat or rising slope on the FR plot.


There are tons of measurements that should dictate performance so Im wondering what is missing. boxiness/openess/spaciousness/soundstage are pretty well the same adjectives so what is really missing in measurements.

I know you will disagree that measurements can be used but is it an experience/usage thing? Acceptance thing?

I'm not disagreeing. I'm simply asking how.

How do you measure: Spaciousness? Size of Soundstage? Warmth? Bloated Sound? Muddiness? Detail? Openess? Boxiness? Accuracy is just a small part of the music equation.

The problem is that there is no general agreement about how a speaker is supposed to measure, and even if there were, all practical speakers fail to measure well.

Speakers have three very audible dimensions for measurement being the absence of nonlinear distortion, their tailored amplitude response, and their directionality, which is of course dependent on frequency.

As a rule, speakers fail to be sonically ideal or even close to ideal in any of these dimensions.

Speakers have fairly disturbingly high nonlinear distortion at various points in their normal ranges of frequencies and SPL level operation. Their amplitude response in any particular direction including on-axis is audibly different from being transparently flat or conformng to some ideal contour. Finally, the whole issue of directionality is a complete mess - there's no agreement about what it should ideally be and any near-ideal compromise you might come up with is usually impractical to implement.

The short answer is that speakers are so non-ideal that they are almost completely unmanagable, if sonic transparency is the goal.

How do you measure: Spaciousness? Size of Soundstage? Warmth? Bloated Sound? Muddiness? Detail? Openess? Boxiness? Accuracy is just a small part of the music equation.

I'm answering this question more than once because it exists in more than just the domain of speakers.

The conditions mentioned above exist first and foremost as characteristics of audible performances (music, dialog, efx, etc.).

If we consider accuracy to be the primary positive value then we don't want an abitrary collection of the above conditions when we listen to speakers. We want a recreation of the weighted collection of the above conditions that existed at the origioinal performance when we listen to speakers.

If we consider personal preference to be the primary value, then we want to allow the listener to have his preferred weighted collection of the above conditions when he listens.

How do you measure: Spaciousness? Size of Soundstage? Warmth? Bloated Sound? Muddiness? Detail? Openess? Boxiness? Accuracy is just a small part of the music equation.

With regard to electronics and players, these qualities are very easy to reproduce with sonic accuracy. All you need is readily achievable levels of nonlinear and linear distortion and dynamic range.

If you wish to manage these properties, then you need a fairly highly-divided recording, something far more channelized than 7.1 or 11.1 or whatever. You also need some fairly sophisticated signal processing.

For example, just about everything we call "the acoustic" will probably have to be fairly carefully synthesized.

I do not think that the subtleties of a speaker in reproducing, for example, the brassy sound of a trumpet or the reediness of an oboe, can necessarily be demonstrated by the speaker's FR. I could be completely wrong.

I think that you can measure a speaker's spectral output when playing the sound of a trumpet playing a middle C (or whatever) and I think that you can demonstrate spectral differences in the sound that different speakers reproduce in this scenario. And this, of course, can be compared to the actual spectral analysis of the trumpet's sound, itself. So, a particular speaker may be really good at replicating a trumpet's sound, but poor at reproducing that exact same note on a piano. Or vice versa. And it is these subtleties that provide a speaker with its characteristic sound, and it is these differences between speakers that we notice and listen for when we audition speakers.

When a sound (signal) is sent to a speaker, a speaker doesn't reproduce it "purely". It has to go through a crossover, and how the signal is parsed and affected by the crossover's design is very important. And then the signals are sent on to the drivers, each of which will have a specific response that is dictated by the drivers' specific design and the electrical and physical characteristics that are engineered into it. And you of course have the speaker's physical characteristics like the size and shape of the cabinet, the drivers' relative location and relationship to the front baffle, the distance between the drivers, etc., etc., etc.. The net result is the "sound" that a particular speaker demonstrates. I think that this is all very complex and way too much so to be demonstrated simply by an FR plot.

I do not think that the subtleties of a speaker in reproducing, for example, the brassy sound of a trumpet or the reediness of an oboe, can necessarily be demonstrated by the speaker's FR. I could be completely wrong.

The first problem with the sound of speakers in a room is that you can break the frequency response of the speaker into two major groups that are both very important. One is the direct response of the speaker, and the second group is the room's response to the speaker as it gets back to your ears, or the indirect sound.

Some modern speakers have good enough on-axis response that their direct response is a pretty good facsimile of the origional sound as recorded.

I have to make the distinction "as recorded" because recordings are probably a lot less accurate than you may think they are. "Every speaker is a hostage of the recordings that people play through it."

Secondly, the room has very profound effects on your perceptions of the speaker's sound. "Every speaker is a hostage of the room you listen to it in." Most of these effects come through what I called the indirect sound.

In order to obtain so-called accurate reproduction, both the direct and indirect sound must be controlled very carefully.

It is a fact that if you move speakers around between different rooms, you may well find that what you think of as the characteristic sound of the speaker actually stays pretty much the same for all reasonably good speakers you put into that room.

Just as surely as very few people actually do what I would consider to be a good listening test, very few people know what their speakers would sound like in an ideal room for that speaker.

I think that one reasonable goal for speaker designers would be to make the sound of a speaker as independent of the room as is reasonably possible.

Play each sine wave one band after another from say the average norm 80Hz to 1 KHz keep it within ±2db 70dbc 0db if its above a bit okay if its down below 2db rise it up. It shouldn’t take you no more than a few days to go though the centre channel.

Some tones will go right above the level other will have less at the listening position. Not sure about Tom Holman con this Auduesy or how ever its spelt I don’t care not interested it sounds like a load of hot air.

You’d need an EQ to cover a multitude of frequencies not sure how many you can get out of the Behranger FEQ2496 I’ll have to do a count but it could be looped together to produce more than 31 band EQ or even 200 300 400 and more I’m guessing of coarse I’ll have to do a count tonight.

Test the first frequency sine wave at 60Hz or what ever on the centre and move up one step at time while using a sound generator like the REW one or other TrueRTA if the frequency goes over +2db lower it keep all the Q settings at the narrowest and go though until each tone will be within ±2db.

I did this for the sub bass the a few nights back and wow smoother than before, I got feed up with that sodden frequency sweep, I’ll use that once I’m sure the range is within ±2db.

The first problem with the sound of speakers in a room is that you can break the frequency response of the speaker into two major groups that are both very important. One is the direct response of the speaker, and the second group is the room's response to the speaker as it gets back to your ears, or the indirect sound.Right. I didn't even mention the room and the speaker's physical relationship to the room. I was focusing mainly on what happens as the signal goes through the speaker and on to reproduction and the differences that are incurred or occur during that process. Once in-room, you can obviously not separate the two; the speaker's direct response and it's in-room response.


It is a fact that if you move speakers around between different rooms, you may well find that what you think of as the characteristic sound of the speaker actually stays pretty much the same for all reasonably good speakers you put into that room.So, is the "characteristic sound" of a speaker just too complicated, with too many variables, to really be tangibly measured? I assume that speakers still have a "characteristic sound" in an anechoic environment, as well.


(I added some more to my post, above, btw.)

FWIW, Floyd Toole addresses this issue (measurements vs. listening) over and over in his book. He makes recommendations for a measurement suite that correlates well with the listening preferences from statistically-valid tests. Well worth reading.
Sound Reproduction: Loudspeakers and Rooms
by Floyd E. Toole. Published by Focal Press, an imprint of Elsevier (Oxford, England, UK; www.elsevier.com). 2008. Paperback, 550 pages, ISBN 978-0240520094. $49.95.

So, is the "characteristic sound" of a speaker just too complicated, with too many variables, to really be tangibly measured? I assume that speakers still have a "characteristic sound" in an anechoic environment, as well.


I don't think that the problem is to complicated to ever be solved, but it is clear to me that it isn't being adequately solved these days.

For example, estimating the direct and indirect response of a speaker in a room is *just* a mater of convolving its frequency response on every axis, including of course the room, and adding the responses together.

I think that the answer might be best presented experientially. IOW you type a rough description of a room, presse a button for processing, and then you download a file that you play though certain headphones or earphones and you hear a pretty fair approximation of what the speaker would sound like in your listening room.

I do not think that the subtleties of a speaker in reproducing, for example, the brassy sound of a trumpet or the reediness of an oboe, can necessarily be demonstrated by the speaker's FR. I could be completely wrong.

I think that you can measure a speaker's spectral output when playing the sound of a trumpet playing a middle C (or whatever) and I think that you can demonstrate spectral differences in the sound that different speakers reproduce in this scenario. And this, of course, can be compared to the actual spectral analysis of the trumpet's sound, itself. So, a particular speaker may be really good at replicating a trumpet's sound, but poor at reproducing that exact same note on a piano. Or vice versa. And it is these subtleties that provide a speaker with its characteristic sound, and it is these differences between speakers that we notice and listen for when we audition speakers.

These are matters of timbre: harmonics (a frequency domain phenomenon) and attack (time domain).


When a sound (signal) is sent to a speaker, a speaker doesn't reproduce it "purely". It has to go through a crossover, and how the signal is parsed and affected by the crossover's design is very important. And then the signals are sent on to the drivers, each of which will have a specific response that is dictated by the drivers' specific design and the electrical and physical characteristics that are engineered into it. And you of course have the speaker's physical characteristics like the size and shape of the cabinet, the drivers' relative location and relationship to the front baffle, the distance between the drivers, etc., etc., etc.. The net result is the "sound" that a particular speaker demonstrates. I think that this is all very complex and way too much so to be demonstrated simply by an FR plot.

I don't see many 'simple' FR plots , or speaker measurement reports, these days.

I'm not disagreeing. I'm simply asking how.

Got it, Me too :D

The short answer is that speakers are so non-ideal that they are almost completely unmanagable, if sonic transparency is the goal.

Still measurements tell a great deal about the speaker. I think we should not worry about debate if speakers are truely accurate instead I want to understand that we can define a speakers by all the existing measurements that can be taken.

I also will agree with the opinion that we do not have enough measurements done on speakers.

It is a fact that if you move speakers around between different rooms, you may well find that what you think of as the characteristic sound of the speaker actually stays pretty much the same for all reasonably good speakers you put into that room

.
.
.

I think that one reasonable goal for speaker designers would be to make the sound of a speaker as independent of the room as is reasonably possible.


That's the inevitable compromise....as it would be impossible for even that speaker to sound accurate in 'every' room....

The goal of NRC-type design is to measure well under well-defined conditions (anaechoic) where 'measure well' also correlates to blind-comparison listener preference*. At the very least a truly 'versatile' loudspeaker would require that, plus some dynamic room-adaptive EQ function to even begin to consistently approach accurate performance at the listening position in different rooms (NB that B&O and a few others make powered loudspeakers with built-in room correction technology...)


*It's interesting in Toole's book that he mention, a few times, a data trend that recording engineers and certain other putative 'golden ears' do NOT prefer the usual winners in Toole/Olive-style blind tests...suggesting perhaps that such listeners are listening for different things, even compared to listeners trained to hear known distortions, as they are in Toole/Olive trials..

Right. I didn't even mention the room and the speaker's physical relationship to the room. I was focusing mainly on what happens as the signal goes through the speaker and on to reproduction and the differences that are incurred or occur during that process. Once in-room, you can obviously not separate the two; the speaker's direct response and it's in-room response.


So, is the "characteristic sound" of a speaker just too complicated, with too many variables, to really be tangibly measured? I assume that speakers still have a "characteristic sound" in an anechoic environment, as well.


(I added some more to my post, above, btw.)


Sound is not that complex, it seems we always want to make it complex because it leaves the debate open forever.

If you listen to driver after driver after driver you will know one thing.....you can know the sound of the driver from accurate measurements.

Open baffle seems to lend itself to more accurate sound reporduction because no box is coloring the sound but box will add things like a better low end to the sound of a speaker, etc.

Im still not sure about a 'wind instrument' vs 'string instrument' difference being unmeasureable or that its really even part of this discussion.

One of the reasons you will hear a timbre difference in a given instrument between two speakers with a similar FR response is this:

Speakers are not linear in their power response over frequency.

In short, if you were to take a FR plot at an average 60 dB SPL, then use an EQ to get the response ruler-flat at 60 dB, you will theoretically be able to reproduce this response. However, if you increase the drive level to get a 90 dB SPL output, the response will no longer be ruler-flat. Since an instrument with lots of harmonics (oboe, distorted electric guitar, cymbals, snare drum, etc) relies on the relative intensity at each of the harmonic frequencies to achieve its unique sound, this lack of output linearity vs. frequency can create "strange" sound quality - even with a speaker system and room that ostensibly measure dead-flat (at one given reference level).

Sound is not that complex, it seems we always want to make it complex because it leaves the debate open forever.

No, it's not terribly complex, but it's not uni-dimensional, either. To really understand what is happening requires far more degrees of measurement freedom than most buyers (or even designers) are willing to (or capable of) assimilating the data for.

Im still not sure about a 'wind instrument' vs 'string instrument' difference being unmeasureable or that its really even part of this discussion.

See above - the difference in overtone series' and their relative levels is a HUGE part of the way we perceive timbre in instruments.

I love measurements. I am an engineering type. But I also know different people require different clues to what is accurate. Different people can "accept" an overall imbalance and their brain says, "this room is flat, OK" Well trained musicians go crazy if the frequency (old record speed) is off a hair. I am sensitive to whatever clues give a spatial image. I would not know an sharp from a flat by itself.

We don't really know what to measure, as we don't quite know how we process hearing.
Even the Hafler test I mentioned in the DIY threads, just compare in vs out, can be easily passed by several amps, then lined up and they sound different. .oo1 THD, IMD, TIM, whatever, I don't know, but different.

I know my wife can hear things our dogs can't. She can walk by a set of speakers, no, no, no, OK, no, no. If the upper reaches of a trumpet are just slightly off, pow, gives her a headache.

I can't explain why some amps make one of my Julian Bream recordings sound like metal strings. And I mean some really high end ones. I can't say what makes hand-clapping on Clapton Unplugged just plain wrong on many speakers. All I know is I hear differences I can't measure, so that means I am not measuring what is important! It was not too long ago when we learned to measure TIM. Now we have amps that slew to 50K so that is no longer a problem. What's next?

All I know is I hear differences I can't measure, so that means I am not measuring what is important! It was not too long ago when we learned to measure TIM. Now we have amps that slew to 50K so that is no longer a problem. What's next?

Right, that is the question of what needs to be measured so that we can take the data and make it easier to choose.

I also assume this is more so for Music then HT, I believe measurements already can dictate the performance of speakers in HT. Movie playback seldom requires someone trying to pick the right timbre.

One of the reasons you will hear a timbre difference in a given instrument between two speakers with a similar FR response is this:

Speakers are not linear in their power response over frequency.

In short, if you were to take a FR plot at an average 60 dB SPL, then use an EQ to get the response ruler-flat at 60 dB, you will theoretically be able to reproduce this response. However, if you increase the drive level to get a 90 dB SPL output, the response will no longer be ruler-flat. Since an instrument with lots of harmonics (oboe, distorted electric guitar, cymbals, snare drum, etc) relies on the relative intensity at each of the harmonic frequencies to achieve its unique sound, this lack of output linearity vs. frequency can create "strange" sound quality - even with a speaker system and room that ostensibly measure dead-flat (at one given reference level).

This is where I believe we should be measuring across many levels including the dynamics of the performance. We also have linearity specs of drivers so its not something that is realy unknown.


I guess the amount of time and the end results being only accepted by a fraction of the people out there wouldnt be worth the bother?

This is where I believe we should be measuring across many levels including the dynamics of the performance. We also have linearity specs of drivers so its not something that is realy unknown.

The trouble is that getting it right would require performing a series of FR plots at 1-dB intervals across the entire dynamic range. Then, you would have to devise an EQing system that would effectively de-convolve the resulting response matrix. Very, very possible, but it would take lots of time and effort to develop the system.

I guess the amount of time and the end results being only accepted by a fraction of the people out there wouldnt be worth the bother?

I'm not sure about that. Anyone who listens critically when they are making a purchase would be able to tell the difference - expecially so if they are given source material to listen to that accentuates the linearity/frequency correlation. I should probably stop posting now and start doing the programming....

I used to think most people can hear the difference, just won't pay for it. Unfortunately, we have a generation now who does not even know what music sounds like, so even if they cared, they don't know what to buy. All they have heard is iPods and 60 Hz single tone base distortion generator ( car woofers). Invest in hearing aid companies. The entire post X generation will be needing them!

Market is a valid point. Everything seems to be reduced to a single sound byte. We can't expect to explain why .001% filth harmonic sounds like nails on a chalkboard and 5% second harmonic sounds like melted butter. Oops, I forget, we don't have chalkboards anymore so GenX does not know what they sound like! We only have the squeek from whiteboards. Must be why almost all the traditional "stereo stores" have closed up.
If it is good for music, it will be good for HT. IMHO.

I guess the amount of time and the end results being only accepted by a fraction of the people out there wouldnt be worth the bother?
If the end results are simply measurements giving more reliable evaluations, I think it would not be worth the bother. We'd need to find some more productive use of the measures and calculations that leads to better sound or cheaper fabrication.

I also assume this is more so for Music then HT, I believe measurements already can dictate the performance of speakers in HT. Movie playback seldom requires someone trying to pick the right timbre.

I agree. Most of the time and money I've spent on my system has revolved around 2-channel music. If all I ever used my system for was HT, I would have been happy many thousands of $ ago! :cool:

We can do all sorts of in room or outside measurements to accurately define a speaker. Correct?

To a certain extent. We can get a pretty good idea.


Then why do people still ignore all those measurements and use "How does it sound to you" Mantra.

Because what sounds one way to one person does not sound the same to the next. Nobody has perfect hearing.


Does accurate speaker equate to best measured speaker? (Flat FR plot, etc)

If the goal is accuracy for some people then why don't they use the measurements instead of the flawed "How does it sound to you" appoarch?Yes that would be the correct way. With questions like this it looks as though you are looking for an arguement, one I don't have. Most things
a person is cabable to hear for example, probobly their arguement whether it sounds "accurate" or not, but accurate would be the incorrect term to use since terms like these are used to defined examples that are tested using a controled method.


How it sounds to them has little to do with accuracy most times. I have to believe people just think there is a missing measurement out there and therefore they have to listen.I agree, they will not get far reading nothing besides frequency response measurements.


Obviously, Im assuming accurate measurements....yes manufacturers can lie!! Lets assume a speaker has the measurements you want. Why not buy it based on those?

Some people are sensitive to speaker timbre, different styles of tweeters, other factors.

I agree. Most of the time and money I've spent on my system has revolved around 2-channel music. If all I ever used my system for was HT, I would have been happy many thousands of $ ago! :cool:

I definitely agree that music has its own world of distinct sounds and I honestly do not believe the sceintific idea of "accuracy" is something many people are really looking for. They are actually looking for the perfect sound for just them that is probably inaccurate and they have no idea how that sounds looks in measurements. They also do not want to spend all the time learning about the measurements. I understand all that and obviously there is no way to measure and produce data for people that go down this path.

I still think most people can start to realize Warm has a defined measurement, Detailed/bright as a defined measurement, Soundstage has a defined meaurement and we are closer then many think to having enough measurements that most people could simply use them to narrow down the speakers.

If the end results are simply measurements giving more reliable evaluations, I think it would not be worth the bother. We'd need to find some more productive use of the measures and calculations that leads to better sound or cheaper fabrication.

That is a good point but on the other hand it would create a little less subjectivity in the speaker world. I think subjectivity is a bad good thing because people are mostly confused. Pure Data that simply states the speaker performance would go a long way with people just looking for speakers. The die hard people are always going to touch/listen, ignore the data and Im cool with that.

.....

Yes that would be the correct way. With questions like this it looks as though you are looking for an arguement, one I don't have. Most things
a person is cabable to hear for example, probobly their arguement whether it sounds "accurate" or not, but accurate would be the incorrect term to use since terms like these are used to defined examples that are tested using a controled method.


Not looking to argue...serious. I believe we can or should be defining speakers better but that goes against the "go listen" crowd's mantra.


Some people are sensitive to speaker timbre, different styles of tweeters, other factors.

I agree but again, the question still remains why can that have a defined measurement too. I do know the partial answer would simply be "Its a waste of time" ';)

I still think most people can start to realize Warm has a defined measurement, Detailed/bright as a defined measurement, Soundstage has a defined meaurement and we are closer then many think to having enough measurements that most people could simply use them to narrow down the speakers.


One thing that I think is missing in this discussion, is component synergy. Since a system works as a unit, 'graphs and measuremnt' of individual components don't tell the whole story.

One thing Penn, I sense that you believe that all one has to do learn to read charts, and specs, and that is all you need to choose quality audio gear. I think that that information can be helpful, and it can even tell you some things about audio gear that you can't get from just listening. But I also feel that listening is a big part of the equation as well. It's not either/or. But both.

I agree. Most of the time and money I've spent on my system has revolved around 2-channel music. If all I ever used my system for was HT, I would have been happy many thousands of $ ago! :cool:I am not trying to be incendiary but, perhaps, it is because you are chasing a two-channel rainbow in an effort to achieve what multichannel can do. :)

I am not trying to be incendiary but, perhaps, it is because you are chasing a two-channel rainbow in an effort to achieve what multichannel can do. :)

No. :p

One thing that I think is missing in this discussion, is component synergy. Since a system works as a unit, 'graphs and measuremnt' of individual components don't tell the whole story.


I agree the do not tell the whole story, rooms, equipment,etc is a whole difference beast and do not want to get into this discussion ;)

Could we just assume that the "system" for this discussion has little impact on the sound....an almost perfect hypothetical beast.





One thing Penn, I sense that you believe that all one has to do learn to read charts, and specs, and that is all you need to choose quality audio gear. I think that that information can be helpful, and it can even tell you some things about audio gear that you can't get from just listening. But I also feel that listening is a big part of the equation as well. It's not either/or. But both.


That is what I would like to believe...today that is not true. Over the past 2 years I have moved towards this belief more and more because of everything I learn on the DIY side.

DIYers do not go out searching for the right sound, instead they design and build their sound so I guess we have more faith in whatever measurements we have on hand before we build.

We can listen to individual drivers to get a sense of what we like then we seek expert advice, we search for DIY success stories. We then build the best box we can and either go active or passive in our crossovers. The end goal is usually a measurements party and then we find out how it actually sounds. ;)

So in the end we let the driver specs/measurements, Box design design/specs/measurements, crossover software/measurements and so on lead us to the final product. Its the completely alternate appoarch.

If a speaker measures pretty flat on axis and somewhat good off axis (depending on your needs) at different SPLs, shows very little distortion even at higher SPLs, has a nice impedance curve then its going to be a good speaker and should reproduce any sound properly.
There are other measurements involved too but Im not the expert to tell what those are.

Now how can that speaker actually sound bad?

I am not trying to be incendiary but, perhaps, it is because you are chasing a two-channel rainbow in an effort to achieve what multichannel can do. :)

Im actually all for someon chasing the rainbow!!! :D If it creates huge passion and incredible happiness Im all for it.

My points here actually do not apply to those people. I believe there are many ways to build a great sound system and I believe people should be able to do it many ways so if they have the proper data in front of them some people can make pretty good decisions.

The trouble is that getting it right would require performing a series of FR plots at 1-dB intervals across the entire dynamic range. Then, you would have to devise an EQing system that would effectively de-convolve the resulting response matrix. Very, very possible, but it would take lots of time and effort to develop the system.


A goodly number of speaker plots have been published, taken on 10 dB intervals. The differences are there, but not terribly strong at moderate levels. It would probably be alright to interpopolate the 1 dB values from the 10 dB values, except near the point where the speaker is being seriously taxed.

More signficant are the differences between on-axis response, and the various off-axis responses.

We can do all sorts of in room or outside measurements to accurately define a speaker. Correct?

We can accurately define the specific measurement and the limitations/tolerances of that measurement.


Then why do people still ignore all those measurements and use "How does it sound to you" Mantra.

Why do we still have to ask how a car rides or handles? We have terms and metrics, but the end perception comes from the contribution of many of these. More importantly, no one agrees what the best balance of contributing factors is, and designers will argue as to what each factor really contributes to the perception/use.


Does accurate speaker equate to best measured speaker? (Flat FR plot, etc)

Not once you look at the spacial radiation of a loudspeaker (easy to do these days with $1-2k in testing). So what's the correct radiation characteristic? :rolleyes:


If the goal is accuracy for some people then why don't they use the measurements instead of the flawed "How does it sound to you" appoarch?

How it sounds to them has little to do with accuracy most times. I have to believe people just think there is a missing measurement out there and therefore they have to listen.

The real problem is that it's REALLY hard to get enough measurements on one display, while also explaining the interpretation of all of those and how they interact with each other. There is no ideal for each measurement, as most all of them need the context of other measurements to be significant. Even then we still have nuances of the correlation between the measurements and listening, not the least of which is how they interact with your room.

When I see measurements or take a series of measurements and and know a bit about the operation of the speaker I can make some pretty good estimations of what strengths and weaknesses a design are likely to be observed. There will still be a few unquantified details that aren't obvious unless you get into part-wise measurements, and in the end you still have to listen to do a final check that nothing was overlooked.


There are tons of measurements that should dictate performance so Im wondering what is missing. boxiness/openess/spaciousness/soundstage are pretty well the same adjectives so what is really missing in measurements.

I know you will disagree that measurements can be used but is it an experience/usage thing? Acceptance thing?

It's both a matter of agreement and who it is useful for. 9 times out of 10 it's more useful for your competitors than to any potential customer. :rolleyes:

If the goal is accuracy for some people then why don't they use the measurements instead of the flawed "How does it sound to you" appoarch?

The issue is personal taste. Accuracy in its purest form is not always exactly euphonic.

Sound is not that complex............Oh. OK.


With questions like this it looks as though you are looking for an arguement................No, not penngray!

Wow, great thread. :D

Then why do people still ignore all those measurements and use "How does it sound to you" Mantra.

There are two types of people - those who understand science, and those who do not. Worse, often those who do not understand science prefer to believe that somehow "science" has not yet learned how to measure all aspects of audio. This is the root of all cable arguments. :D

Everything that affects audio fidelity has been fully understood for at least 50 years if not longer. Yes, vendors leave out important data and they also lie by averaging speaker response etc. They do it so often it's become the norm. This does not mean specs cannot give us 100 percent of what is needed.

As for speakers specifically, the speaker with the flattest on-axis response and lowest distortion is the one to buy. Sadly, many people prefer speakers that are intentionally skewed to sound "better." One common trick is to add an intentional dip in the harshness range around 2 to 4 KHz. Speakers like that sound "smooth and airy" and non-fatiguing. Me, all I want is flat, with low distortion. Good luck finding any speaker company that offers distortion data!

--Ethan

Worse, often those who do not understand science prefer to believe that somehow "science" has not yet learned how to measure all aspects of audio. This is the root of all cable arguments.With all due respect, measuring the electrical characteristics of cables and "measuring speakers" are two completely different things.


This does not mean specs cannot give us 100 percent of what is needed.Could you summarize for me all the data one would have to collect in order to fully "understand" a speaker's performance just by looking at the collected info? If it IS all collectable, I am sure that given enough speakers to listen to and enough data to comparatively analyze, one would be able to recognize trends and even predict a speakers "sound" based simply upon said data.

But are you suggesting that we should be able to BUY a speaker based simply upon a stack of data collected on said speaker? I do not care how much data one can collect and analyze, your ears can collect and analyze much more, and much more quickly. And can also, yes, subjectively, weed out what is the unnecessary data. It is the net sum of all that data that would dictate how a speaker sounds, and your ears are much better at deciphering that, subjectively, than your brain would be at analyzing it, objectively.

And speaker preference IS subjective. Always will be. Given the opportunity, buying a speaker without listening, first, is idiotic, IMO, and I do not know anyone who would recommend that. No matter how much "science" one had at hand.

... those who do not understand science prefer to believe that somehow "science" has not yet learned how to measure all aspects of audio.
The contrary of this, which you seem to be urging on us, is not a scientific theory, since it's not susceptible to evidence or counterevidence. So this is just an invitation to a debate on faith in science.

Good luck finding any speaker company that offers distortion data!

In the promotional material on their website for their M2 bookshelf speaker, Axiom links to a review giving a graph of "THD+N", here (http://www.soundstagemagazine.com/measurements/axiom_m2i/), among other specs.

There are two types of people - those who understand science, and those who do not.

Hate to undermine the ego there partner, but knowing or not knowing the science WILL NOT dictate whether that person likes the way a speaker sounds or not. Except in certain rare cases, when a person feels his head getting torn off by a set of Genelecs but revels in "the accuracy.":rolleyes:

Worse, often those who do not understand science prefer to believe that somehow "science" has not yet learned how to measure all aspects of audio. This is the root of all cable arguments. :D

It's easy to measure a speaker, not so much to measure preference. People sort of know it when they're exposed to it.

Everything that affects audio fidelity has been fully understood for at least 50 years if not longer. Yes, vendors leave out important data and they also lie by averaging speaker response etc. They do it so often it's become the norm. This does not mean specs cannot give us 100 percent of what is needed.

Good quality companies measure anechoically, because they cannot accurately predict everyone's environments. I would think you would appreciate having this standard as a platform to launch from. It's kinda irrelevant what poor companies do, because they probably shouldn't be under consideration anyway for any number of other reasons.

As for speakers specifically, the speaker with the flattest on-axis response and lowest distortion is the one to buy. Sadly, many people prefer speakers that are intentionally skewed to sound "better." One common trick is to add an intentional dip in the harshness range around 2 to 4 KHz. Speakers like that sound "smooth and airy" and non-fatiguing. Me, all I want is flat, with low distortion. Good luck finding any speaker company that offers distortion data!

--Ethan

Selling speakers is primarily a business decision, not whether Random Enthusiast A out of thousands finds the sound disagreeable. The trick is finding the most accurate approach, while maintaining a degree of mass appeal. Tough line to straddle, I'll grant you.

I don't think we have that information at this time miles with respect to speakers. From what I understand, efforts are being made at Harmon to identify those measurements apart from FR that also correlate well with listener preference. Properly weighted for importance and signficance, I can imagine there is more than one 'optimum' combination of parameters that listeners would find equally pleasing. Factor analysis...resolution mapping are but some of the techniques that in the hands of a competent mathematician can yield useful nuggets of information. On a related but tangential area, have you read Numerati?

From what I understand, efforts are being made at Harmon to identify those measurements apart from FR that also correlate well with listener preference. Properly weighted for importance and signficance, I can imagine there is more than one 'optimum' combination of parameters that listeners would find equally pleasing. Factor analysis...resolution mapping are but some of the techniques that in the hands of a competent mathematician can yield useful nuggets of information.I think that this MAY be a worthy endeavor, but what it will probably reveal is subsets of different listener preferences. Still, I do not think anyone could or should buy a speaker based upon data. Are we going to have some sort of eHarmony-type speaker personality match? :D


On a related but tangential area, have you read Numerati?No

Understanding what's important, and I'm not saying that's the case now, can provide a useful means to develop a better short list. It doesn't have to be a means solely in of itself to make a choice without listening, but it can especially for those people who are looking for their first real speakers and are now presently buying on the strength(s) of recommendation(s).
I do recommend the book though.

Understanding what's important, and I'm not saying that's the case now, can provide a useful means to develop a better short list.

This is a standard product development technique. It's called QFD, and it's part of the Six Sigma toolkit. Of course, setting the input data for the filtering process is the really sticky part....

Understanding what's important, and I'm not saying that's the case now, can provide a useful means to develop a better short list. It doesn't have to be a means solely in of itself to make a choice without listening, but it can especially for those people who are looking for their first real speakers and are now presently buying on the strength(s) of recommendation(s).So, how would it "work"? You'd be able to find out what speakers may have a similar sound to a reference speaker that you DO prefer? I can see how it could be useful in providing info such as "if you liked speaker A, then you should like Speaker B (or not like Speaker C)". But, otherwise, how is one going to express their preferences in a detailed and analytical enough way to make the data useful or assignable? How many ways are there for a person to describe the sound they seek, anyway? And if one goes about things so strictly, they may likely miss out on a speaker that they may have been absolutely thrilled with. Why rule a speaker out based simply upon what is essentially marketing data? We don't always know ourselves as well as we think we do and our preferences can easily be swayed. I would prefer to be enlightened instead of pigeon-holed. Ears wide open.

I’m have a strong appreciation for technology, engineering and science, so when I first got into this hobby I took it upon myself to learn to the best of my ability what all of the published measurements meant. I planned to use this information to narrow down my selection of speakers to only those with excellent measured performance (flat on and off-axis FR, clean waterfall, low distortion, etc.). I wanted objectively verifiable accuracy.

Logically thinking, I figured that if the speaker was doing everything that it was supposed to within a tight tolerance, that the speaker would be “telling it how it is”. And if something sounded bad when played through said speakers, then it was clearly just a bad recording.

However, upon listening to several “accurate” speakers (Aerial, Revel, KEF Reference, Paradigm Sig) I found that all had distinct characteristics that didn’t coincide with the published measurements, at least to these ears. The Aerials sounded dull and boxy, the Revels sounded dry and clinical, the KEFs sounded bright and fatiguing, and while the Paradigms sounded the best of all of them, they could sound a bit forward and “shouty”. Look at the measurements published in Stereophile or Soundstage on any of these speakers and you would think that they would all sound pretty similar. They didn’t.

So while I used to narrow down my choices by using published measurements, I've heard too many speakers that measure well, but don't sound "right" to rely on the measurements as a selection tool. Today I find it more effective to listen to every speaker in my price range that I have access to, and refer to the measurements just to be sure that they are doing anything too wonky. Even if a speaker doesn't measure well, rather than rule it out, I would still audition it and listen for problems in the areas that the graphs point to.

Now I’m not throwing the baby out with the bathwater here. I do believe that the commonly published measurements can tell us useful information about a speaker's performance and I use Room EQ Wizard and PEQ to flatten the in-room response of my system, but I've also found that there can be things just as, or more important to sound quality than what is shown in a Stereophile / Soundstage graph. Things I hear that don't correlate with the published graphs or my own measurements.

For example, I owned some Focal speakers for a while. When I measured the response of these speaker in my listening room, there was a 3dB dip in the cross-over region that spanned from about 2KHz to 4KHz. You would think that this sort of measured response would give the speaker a somewhat laid-back, relaxed presentation. Yet, my primary concern with these speakers after owning them for a while was upper midrange / lower treble harshness. The subjective and the objective didn’t add up. In fact they seem to contradict one another.

I’m not saying that what I heard in the Focal’s cannot be measured, rather that the needed measurements are not readily available. Some sort of detailed distortion measurements, not just THD would probably be a good start.

For the time being, it is just much easier to listen.

I think the first step is to introduce perceptual models. Right now, people publish the standard "linear systems" kinds of measurements like frequency response, because it's a well-understood first try. IMO if you want to get much beyond this, and correlate the perceived performance of an audio system with some measurements, you need to model the listener as well. This seems to be an active area of research.

The basic idea is this: the perceptual models (as used in lossy compression) are good at separating audible parts of the signal from inaudible parts. If you look at what the model thinks the difference is between the original and the playback signal, you should get more obviously useful information about the differences than you would looking at the raw data, or linear systems types of plots. This idea is being used in telephony, to automatically evaluate the quality of speech sent over networks using various codecs.

Here's one thought experiment to illustrate the point: take a CD of a tune, and then play the same tune back through a stereo system in a dead room. Record the results at the listening position. Subtract the recorded signal from the original signal on CD, and make an mp3 of the difference. The number of bits in that mp3 should tell you how audibly different the two signals are. If the mp3 is tiny, you've got a good reproduction: other measurements, say total energy of the signal, might show a lot of distortion, but the perceptual model is telling you that it isn't very audible. If the mp3 is long, you're introducing a lot of audible distortion.

I can't explain why some amps make one of my Julian Bream recordings sound like metal strings. And I mean some really high end ones.

Is that because you've measured them and don't see anything in the measurements that would seem to relate to what you hear?

Some 'really high end ones' are really badly designed, and that will show up in measurements.


I can't say what makes hand-clapping on Clapton Unplugged just plain wrong on many speakers.

Porbably the recording itself.

Though someone else, who is more familiar with what can be measured in loudspeakers, might be able to define what is going wrong on those speakers.


All I know is I hear differences I can't measure,


What sorts of parameters have you measured?

I also assume this is more so for Music then HT, I believe measurements already can dictate the performance of speakers in HT. Movie playback seldom requires someone trying to pick the right timbre.

Actually if there's anything humans are acutely sensitive to, it's human speaking voice.

If that isn't 'right', people notice.

With all due respect, measuring the electrical characteristics of cables and "measuring speakers" are two completely different things.


Could you summarize for me all the data one would have to collect in order to fully "understand" a speaker's performance just by looking at the collected info? If it IS all collectable, I am sure that given enough speakers to listen to and enough data to comparatively analyze, one would be able to recognize trends and even predict a speakers "sound" based simply upon said data.

But are you suggesting that we should be able to BUY a speaker based simply upon a stack of data collected on said speaker?

I would suggest that. Because if you do it 'by ear', you are really listening to the loudspeaker + room + biases. And the loudspeaker that sounds good to you in the salon, is by no means guaranteed to sound that way in your home. And if it sounds good in one room of your home, there is no guarantee it will dot the same in another. Nor is there a guarantee it will sound good to you if you could actually hear it in blind comparison. It's a real crapshoot.

Whereas if you are buying based on a comprehensive set of measurements , including anaechoic (where only the loudspeakers is being measured, not the room) you AT LEAST have a baseline of performance. IN otherwords, you can focus more on fixing the ROOM's problemss, not the loudspeaker's.


I do not care how much data one can collect and analyze, your ears can collect and analyze much more, and much more quickly.


So what? They aren't usually collecting it under conditions that make the information reliable for drawing cause/effect conclusions.



And can also, yes, subjectively, weed out what is the unnecessary data.



Are you suggesting we can by force of will, 'weed out' subjective bias, not to mention the effects of the room, on the 'actual' performance of the loudspeaker?

We can't. But we are, fortunately, pretty good at ignoring (or at least acclimating to) what might measure as pretty lousy in-room performance.
Toole's book has some fascinating reports on this.

It is the net sum of all that data that would dictate how a speaker sounds, and your ears are much better at deciphering that, subjectively, than your brain would be at analyzing it, objectively.

That would be nice to see tested, but I'm unaware that anyone has.



And speaker preference IS subjective. Always will be. Given the opportunity, buying a speaker without listening, first, is idiotic, IMO, and I do not know anyone who would recommend that. No matter how much "science" one had at hand.

I see nothing wrong with it -- or at least, nothing worse than buying the totally 'subjective' way -- so long as you understand what you are doing.

I've certainly bought loudspeakers 'sound unheard' and been happy with them.

Hate to undermine the ego there partner, but knowing or not knowing the science WILL NOT dictate whether that person likes the way a speaker sounds or not. Except in certain rare cases, when a person feels his head getting torn off by a set of Genelecs but revels in "the accuracy.":rolleyes:

Nope, it's won't *dictate* anything, but 'the science' does so far, does tell us about some interesting likelihoods. One being that listeners, especially if they are trained to hear typical forms of loudspeaker distortion, DO tend to converge on preference for certain measured performance -- e.g., flat on and off-axis response -- IF they are allowed to FAIRLY compare loudspeakers (that is, double-blind)

It's a probability claim, but all scientific 'facts' are that way.


It's easy to measure a speaker, not so much to measure preference. People sort of know it when they're exposed to it.

It's been done: measured performance of a speaker versus listener rating of the sound quality. Look up the work fo Sean Olive and Floyd Toole, across a few decades of JAES publication. Also look up references to 'Canadian NRC'
in brochures by Paradigm, JBL, and some other loudspeaker manufacturers.

I think that this MAY be a worthy endeavor, but what it will probably reveal is subsets of different listener preferences. Still, I do not think anyone could or should buy a speaker based upon data. Are we going to have some sort of eHarmony-type speaker personality match? :D


Why joke about it? That would be an excellent goal to aspire to, as thre's evidence that there may be at least two broad categories of 'preference convergence'. It would also be a far more efficient process than the typical 'audiophile audition' method of choosing loudspeakers.

Have you actually acquainted yourself with the literature on determinants of loudspeaker preference? I think you'd find it interesting...and perhaps more well-developed than you realize.

Understanding what's important, and I'm not saying that's the case now, can provide a useful means to develop a better short list.

I think it's already possible to do that, thanks to the work done so far: draw up a short list of loudspeakers likely to 'sound good' to many people based on good measured performance. The real limiting factor for this approach is the unavailability of comprehensive and consistent measurement across the specturm of available loudspeakers. The list is 'shorter' than it could perhaps be, because some speakers simply haven't been measured properly yet.

Another confounding factor is that while the speaker itself may be accurate, the room it's in may 'sound' very bad. But that is NOT the loudspeaker's 'fault'.

Personally, in an age of abundant 'analog' and digital means of room correction, I don't want loudspeakers that only suit one room well.

Being able to choose speakers solely by measurements is an interesting concept. And perhaps one worth pursuing.

First you have to agree on a "standard" method to make the measurements. I guess anechoic would be the obvious choice. Not everybody has access to that type of facility.

Then you'd have to decide what needs to be measured, and how it should be displayed. On axis frequency response, off axis frequency response, distortion, impulsive response, resonant behavior, and perhaps other things I can't think of right now would be obvious options.

Then you'd have to figure out what set of measurements resulted in a speaker that you might like the most, and would work in your room with your equipment. No speaker will measure perfectly, so you'd have to learn which set of measurements were most important to you, and what deviation from (perfect, if there is such a thing) were most acceptable to you.

I think most people would agree you want a speaker with flat frequency response both on and off axis. You want one with no distortion, and no resonances. You want one with uniform impedance with zero phase. Good luck getting even one of those things.

As is it right now measurements can at best provide some guidance. Even given a complete set of measurements on every speaker available in my price range I'm not sure I could choose which one I'd like the best. Why? Because I haven't spent countless hours determining which measurements are most important to me. In the end as a consumer it is just easier to go out and listen to as many speakers as possible, then somehow decide which I like the best.

Don't get me wrong. Measurements can provide helpful guidance for consumers, and can be even more useful for designers. Even less than ideal ones such as published in some magazines are useful - but magazines can measure only a small fraction of speakers - you'll never get all makers to agree on a set of measurements, and how to take them. Being able to select speakers based only on measurements is a pipe dream as far as I can see - nothing wrong with dreaming though :)

But are you suggesting that we should be able to BUY a speaker based simply upon a stack of data collected on said speaker?I would suggest that. Because if you do it 'by ear', you are really listening to the loudspeaker + room + biases. And the loudspeaker that sounds good to you in the salon, is by no means guaranteed to sound that way in your home. And if it sounds good in one room of your home, there is no guarantee it will dot the same in another. Nor is there a guarantee it will sound good to you if you could actually hear it in blind comparison. It's a real crapshoot.Obviously, listening in-home is always preferable, but not always possible. Why is actually listening to the speakers a crapshoot? Is it more of a crapshoot than relying upon someone else's crunched data?


I do not care how much data one can collect and analyze, your ears can collect and analyze much more, and much more quickly.So what? They aren't usually collecting it under conditions that make the information reliable for drawing cause/effect conclusions.I'll rely upon MY ears, honed by thousands of years of evolution, to make MY speaker-buying decisions. Seriously, why would you do it any other way?


And can also, yes, subjectively, weed out what is the unnecessary data.Are you suggesting we can by force of will, 'weed out' subjective bias, not to mention the effects of the room, on the 'actual' performance of the loudspeaker?No, I am simply suggesting that we can subjectively choose which speakers sound best to us. Even with the caveats, do you not think that someone should choose their own speakers, subjectively?


It is the net sum of all that data that would dictate how a speaker sounds, and your ears are much better at deciphering that, subjectively, than your brain would be at analyzing it, objectively.That would be nice to see tested, but I'm unaware that anyone has.What would you test? Whether "what I hear" is really "what I hear"?


And speaker preference IS subjective. Always will be. Given the opportunity, buying a speaker without listening, first, is idiotic, IMO, and I do not know anyone who would recommend that. No matter how much "science" one had at hand.I see nothing wrong with it -- or at least, nothing worse than buying the totally 'subjective' way -- so long as you understand what you are doing.

I've certainly bought loudspeakers 'sound unheard' and been happy with them.Do you really think that numbers can somehow be more discerning than our own ears? They are, after all, what we hear with. No matter how many numbers you generate, you cannot rival the audio processing capability of our ears (and brains). A marvelous organ, those ears.


Are we going to have some sort of eHarmony-type speaker personality match? :DWhy joke about it?Because I think it is funny? ;) :o


It would also be a far more efficient process than the typical 'audiophile audition' method of choosing loudspeakers.It would? As would be using a similar method to determine what food I eat or art I buy? Or car? Or clothing? Or what color I paint my house? No, I don't want to spend weeks auditioning speakers. Or do I? I've acknowledged that there may be a way to whittle down the choices somehow, but what if the one speaker I would really like were placed on the "don't listen" list?


Have you actually acquainted yourself with the literature on determinants of loudspeaker preference? I think you'd find it interesting...and perhaps more well-developed than you realize.I realize that a lot of research has been done regarding this. As has similar research been done into what french fries I should prefer and why.



If it were so easy, there wouldn't be so many different speakers for us to choose from. But there are. And I think that that, alone, speaks volumes. Loudly.


And forgive me for being flippant. I just think it is a bit absurd to forgo using your ears to choose speakers.

I would suggest that. Because if you do it 'by ear', you are really listening to the loudspeaker + room + biases. And the loudspeaker that sounds good to you in the salon, is by no means guaranteed to sound that way in your home. And if it sounds good in one room of your home, there is no guarantee it will dot the same in another. Nor is there a guarantee it will sound good to you if you could actually hear it in blind comparison. It's a real crapshoot.

Whereas if you are buying based on a comprehensive set of measurements , including anaechoic (where only the loudspeakers is being measured, not the room) you AT LEAST have a baseline of performance. IN otherwords, you can focus more on fixing the ROOM's problemss, not the loudspeaker's.


To me, timbre is vitally important, yet it has been my experience that the room has little affect on the *timbre* of a speaker from the midrange on up. The room can drastically affect the bass quality, and imagining, and clarity, but rarely timbre.


I've had a number of speakers in my room and listened to them both in the near-field and far-field, and EQ'd them for flat response, and there is still a ot of character to each of them that is independant from the room sound.

I listened to my current speakers at the dealer before he allowed my to bring them home. The dealer's showroom is highly reflective and reverberant, with exposed hardwood floors and exposed brick walls. Was the sound good in this environment? No. BUT, what I heard was excessive reverberation, not excessive brightness, or a change in the tone of the speaker. The room made for a jumbled sound, but I could still seperate the speaker from what the room was doing. And when I brought the speakers home, to my acoustically treated room, the tonal character of the speaker remained while the room treatments cut down on the clutter.

I believe that we, as humans, give priority to the direct sound and tend to filter out the room to a large degree.

Are we going to have some sort of eHarmony-type speaker personality match? :D


That is actually an interesting concept. Choosing a speaker has some similarities to choosing a spouse. Maybe. There is no right answer for everybody. Apparently an eHarmony type of matching can be somewhat successful, and least in narrowing down potentials. I don't know.

A client could answer a bunch of questions, and some cleverly designed software could suggest options - of course you'd have to decide what questions to ask the speakers ;) Could be a money maker.

Choosing a speaker has some similarities to choosing a spouse. Maybe. There is no right answer for everybody. Apparently an eHarmony type of matching can be somewhat successful, and least in narrowing down potentials. I don't know.Sure, it may narrow down potentials. Roughly. But you might completely miss your soulmate, too. :(

Choosing a speaker is a hell of a lot easier, btw. As is changing your mind. :eek:

Choosing a speaker is a hell of a lot easier, btw.

I agree. And it is much easier to upgrade as technology advances.
I expect to have one spouse, but many speakers. Perhaps if we
were as tolerant of speakers as we were with spouses we would
not want to upgrade. "Sure it is not perfect, but it is pretty good
and I'm happy enough" - nah - works for spouse, not for speakers.

In a way this site works as a freebie eHarmony for speakers. People
post asking what they should buy. Others ask what they are looking for and how much they are willing to spend. Suggestions are made based on that input. In the end the consumer is either silly/stupid/lazy and just goes with the recommendation, or they take the advice to go out and listen to the most likely options. Thank god I did not take my mothers and sisters suggestion on who I should select for a spouse. My dad and brothers on the other hand have made useful suggestions on speakers I should consider.

It would? As would be using a similar method to determine what food I eat or art I buy? Or car? Or clothing? Or what color I paint my house? No, I don't want to spend weeks auditioning speakers. Or do I? I've acknowledged that there may be a way to whittle down the choices somehow, but what if the one speaker I would really like were placed on the "don't listen" list?


Bingo! Speakers are built based on science and engineering, but it is humans that they have to please. The flaw in buying based of measurements is that there are so many variables that measurements don't account for. Do you have a thing for cymbals? Well then maybe you'd prefer a speaker that emphasises treble. Do you listen to a lot of compressed / bright recordings? Well then maybe you want a speaker that takes the edge off. At best the measurement available in AV mags are a rudementary gauge for quality.

It took me a while to learn this, but I was going about it all backwards. Rather than use measurements to rule out certain speakers due to what seems to be marginal performance, I think measurements are most helpful for pointing out possible things to listen for. Oh, this speaker has a dip in the lower midrange, and a strong resoance in the cabinet in the upper midrange. Well then I'll listen for leanness to vocals and harshness in the upper mids.

The bottom line is that I traded a set of technically near-perfect Revel Studio2's for speakers with a more wavy on-axis FR and a null in the crossover region off-axis, because to these ears, the speakers I have now sound more like real music.

I believe that we, as humans, give priority to the direct sound and tend to filter out the room to a large degree.
That ability is probably derivative from our evolved ability to identify voices regardless of the acoustic environment we happen to hear them in.

I agree. And it is much easier to upgrade as technology advances.
I expect to have one spouse, but many speakers. Perhaps if we
were as tolerant of speakers as we were with spouses we would
not want to upgrade. "Sure it is not perfect, but it is pretty good
and I'm happy enough" - nah - works for spouse, not for speakers.
Now how would upgrade-itis work for spouses?;)

Just for fun, lets get back to the OP's original question

"What part of sound is not measureable?"

I would argue that there is no part of sound that is not measurable. The tough part is what measurements to take, and how to interpret them.

Now how would upgrade-itis work for spouses?;)"I want what he's having."

"What part of sound is not measureable?"

I would argue that there is no part of sound that is not measurable. The tough part is what measurements to take, and how to interpret them.
If when you've taken a measurement, you don't know how to interpret it, in what sense have you measured something? How could you possibly know that every part of sound is measurable when you don't know what to measure nor the meaning of your measures? What is the argument that you would make?

If when you've taken a measurement, you don't know how to interpret it, in what sense have you measured something? How could you possibly know that every part of sound is measurable when you don't know what to measure nor the meaning of your measures? What is the argument that you would make?

I would argue simply that I have taken a measurement, but I don't know how to interpret it. I can easily measure for example pressure, I know that means I have measured the pressure, but I don't know yet how that measurement might effect what I hear, or if it is going to rain or snow. It is easy to measure things without understanding what those measurements mean, if they mean anything at all. Sometimes the measurement can be the challenge, but usually it is knowing what to measure and how to interpret it that is the hardest.

Bingo! Speakers are built based on science and engineering, but it is humans that they have to please. The flaw in buying based of measurements is that there are so many variables that measurements don't account for. Do you have a thing for cymbals? Well then maybe you'd prefer a speaker that emphasises treble.

And maybe it would be easier to know which of them do *that* simply by looking at how they measure, than by listening in a salon. (Regardless of sivadselim's argument, most people don't audition multiple loudspeakers at home).


Do you listen to a lot of compressed / bright recordings? Well then maybe you want a speaker that takes the edge off. At best the measurement available in AV mags are a rudementary gauge for quality.

Some are less rudimentary than others.


It took me a while to learn this, but I was going about it all backwards. Rather than use measurements to rule out certain speakers due to what seems to be marginal performance, I think measurements are most helpful for pointing out possible things to listen for. Oh, this speaker has a dip in the lower midrange, and a strong resoance in the cabinet in the upper midrange. Well then I'll listen for leanness to vocals and harshness in the upper mids.

So far several of you seem to be looking for loudspeakers that have a 'timbre'. My goal in future loudspeaker purchases would be ones that add as little of their own coloration ('timbre') as possible ; I don't want loudspeakers with their own 'sound'; any more than I want CDPs or amps like that. With a 'neutral' loudspeaker I can *make* the sound deviate from flat to *my* specifications, or not, as I choose.


The bottom line is that I traded a set of technically near-perfect Revel Studio2's for speakers with a more wavy on-axis FR and a null in the crossover region off-axis, because to these ears, the speakers I have now sound more like real music.

(shrug) And in another room, they might not. And , too, you haven't really done a fair comparison (blind).

Look, I'll suggest it one more time to a few here: acquaint yourselves with the literature on speaker preference. Or make your way through Toole's 2008 book. Learn about what is known, and not known. One thing that is known is that when people -- including you -- think they are picking loudspeakers just on the basis of their *sound*, they almost certainly are not. So much for the 'my ears uber alles' argument .

I do acknowledge that picking loudspeakers 'sound unheard' is a leap -- we don't have *all* the correlated of good sound to good measurements at hand; research is still ongoing. But I also realize that we aren't where we were ten, twenty years ago. And I think we know enough to at least *play the odds* now...IF one is seeking a 'neutral' loudspeaker.

I can easily measure for example pressure, I know that means I have measured the pressure, but I don't know yet how that measurement might effect what I hear, ...
Yes, but you said every part of sound is measurable, not every part of pressure waves is measurable. If you're measuring pressure, but you have no idea about the perceptual relevance of your measures, you're not measuring any part of sound, much less every part.

What part of sound is not measureable?

The part that determines individual preference. But, on second thought, that's not really part of the sound.

Just for fun, lets get back to the OP's original question

"What part of sound is not measureable?"

I would argue that there is no part of sound that is not measurable. The tough part is what measurements to take, and how to interpret them.

Good point!

Bingo! Speakers are built based on science and engineering, but it is humans that they have to please. The flaw in buying based of measurements is that there are so many variables that measurements don't account for. Do you have a thing for cymbals? Well then maybe you'd prefer a speaker that emphasises treble. Do you listen to a lot of compressed / bright recordings? Well then maybe you want a speaker that takes the edge off. At best the measurement available in AV mags are a rudementary gauge for quality.

It took me a while to learn this, but I was going about it all backwards. Rather than use measurements to rule out certain speakers due to what seems to be marginal performance, I think measurements are most helpful for pointing out possible things to listen for. Oh, this speaker has a dip in the lower midrange, and a strong resoance in the cabinet in the upper midrange. Well then I'll listen for leanness to vocals and harshness in the upper mids.

The bottom line is that I traded a set of technically near-perfect Revel Studio2's for speakers with a more wavy on-axis FR and a null in the crossover region off-axis, because to these ears, the speakers I have now sound more like real music.


There are measurements that will lead you to speakers that will sound right for you.

I really do not think its a flaw for most people. I would say that 90% plus of all people buying speakers have no idea in the first place what they like. They end up buying something that isnt close to good but closer to what they always have been use too.

Again, Im not against people listening, listening and listening to speakers just to find the perfect speaker for them. Its just that they are a small percentage fo the population everyone else just want an accurate speaker that is built properly and plays properly for them in their room at there SPL/listening distance requirements.

I think the hobby as its extremes and that is great but I see the extreme side of the hobby pushing the others down the wrong path. Measurements are extremely important for those who have no idea what good sound actually is.

Put it this way, put 100 people in a room that have never owned a speaker in their life. Let 50 pick speakers on what sounds good to them and let 50 pick speakers based on a wide range of measurements.

Come back 6 months later and find out who kept the speakers and who didnt.

In the end, if we can measure almost everything about sound then we can predict speaker performance that will please the majority of people. There will always be someone out there that doesn't want to do this and that is fine but to call it a flaw is just wrong.

Oh. OK.


No, not penngray!


Thanks for helping the thread :rolleyes:

But lets keep it to a good discussion.

With all due respect, measuring the electrical characteristics of cables and "measuring speakers" are two completely different things.



No cable debates, please!



Could you summarize for me all the data one would have to collect in order to fully "understand" a speaker's performance just by looking at the collected info? If it IS all collectable, I am sure that given enough speakers to listen to and enough data to comparatively analyze, one would be able to recognize trends and even predict a speakers "sound" based simply upon said data.

But are you suggesting that we should be able to BUY a speaker based simply upon a stack of data collected on said speaker? I do not care how much data one can collect and analyze, your ears can collect and analyze much more, and much more quickly. And can also, yes, subjectively, weed out what is the unnecessary data. It is the net sum of all that data that would dictate how a speaker sounds, and your ears are much better at deciphering that, subjectively, than your brain would be at analyzing it, objectively.

And speaker preference IS subjective. Always will be. Given the opportunity, buying a speaker without listening, first, is idiotic, IMO, and I do not know anyone who would recommend that. No matter how much "science" one had at hand.


I think you didnt read about the science part ;) and why do people insist its idiotic? What about the thousands of DIYers and speaker builders out there that build speakers based 100% on the numbers? Are they idiotic?

Our ears/brains are extremely imperfect in terms of what is actually good out there. If people want to improve their listening experience they have to actually train their ears/brain to understand what is a truely accurate sound. Mostly everyone out there refuses to do that and we all just want a SPECIFIC sound. That in itself is doable too....Bright, Warm, Neutral are all easily defined by measurements. in the end I would never tell someone to just go listen to speakers because most of them have zero idea what is truely a good sound and in the end they pick crap speakers that later on they are upgrading again :eek:



Lets leave the idea that its flawed or idiotic to use science to pick speakers out of this discussion, okay?

. ...Suggestions are made based on that input. In the end the consumer is either silly/stupid/lazy and just goes with the recommendation, or they take the advice to go out and listen to the most likely options....


The new customer will fail at picking the right speaker if told to just go listen, Its a very common occurance.

The ones that do use suggestive opinion all the time on here to narrow the speakers down to a couple of popular speakers so the probability of failure is low because those speakers are made to be very pleasing to a broad audience. Again, measurements could dictate what speaker would be choosen for each individual if we had enough data on the individual's tastes.

Again lets leave all the insulting adjectives out of the debate...or get another thread so we can argue about improper room setup, no treatements, bad EQing (those are much bigger flaws then people buying popular speakers ;) ).....then on too car choices, jobs, education, money decision or the many other flawed decisions people make in the world that are much, much worse then buying just a speaker :rolleyes:

The new customer will fail at picking the right speaker if told to just go listen, Its a very common occurance.



Why?

If the customer listens to more than one pair of speakers, isn't the "right" speaker the one they think sounds best?

Why?

If the customer listens to more than one pair of speakers, isn't the "right" speaker the one they think sounds best?

The right speaker is only the right speaker for a little while usually. If they listening long enough they will start to realize its imprefect and they will want to upgrade.

The key here is that they have no idea what sounds the best so telling them to just listen does not help them at all. They need to first learn what sounds good.

This is just the same as learning something like golf. A long time ago one of my good friends (ex-pro, played the US open...just once ;) ) told me that I could go out and hit 10,000 balls and never learn a damn thing. I need to be taught the specifics first then I can actually learn golf.

The same is true for the speaker world. We should be teaching people what truely sounds better...heck, I know it took me a while to appreciate accuracy because like everyone else I never heard it forever I was use to what sounded good to me. It takes time and education but like a golf swing, it takes education and time!

The measurements give us the basis for teaching people what is more accurate sound, if they learn about it and still choose something else. That is okay too but it has to be taught first.

measuring the electrical characteristics of cables and "measuring speakers" are two completely different things.

Yes, cables are trivial, and loudspeakers are very complex. But in the context of this thread, everything that affects the sound of a speaker can be measured.

Could you summarize for me all the data one would have to collect in order to fully "understand" a speaker's performance just by looking at the collected info?

I'm not a speaker expert, so this list may be incomplete:

* On-axis frequency response.

* Off-axis response, not only side to side but also top to bottom, at every angle from 0 to 360 degrees.

* THD, and also IMD at many frequency combinations to assess all drivers and all crossover components. Distortion should also be measured at several output levels from moderately soft to as loud as the speaker can play just short of damage.

* Ringing, again at enough frequencies to assess all drivers and all crossover components.

* Maximum output level versus frequency.

are you suggesting that we should be able to BUY a speaker based simply upon a stack of data collected on said speaker?

If all of the needed data were available, and buyers understood all the data presented, then Probably Yes.

speaker preference IS subjective.

According to Floyd Toole, speakers that measure the best also sound the best to both experienced and inexperienced listeners. Which makes sense, with three exceptions:

1) Inexperienced listeners are often fooled by response deviations that sound pleasing, such as the phony bass boost in some Bose speakers, and the too-common reduction in the harshness range around 2 to 4 KHz that can make a speaker sound "clear and airy" and non-fatiguing.

2) The sound of any speaker is influenced heavily by the room you listen in, and a speaker that sounds good in a store will surely sound different in your home.

3) Related to 1) above, sometimes a speaker sounds good initially, but after living with it a while its flaws become more apparent. I think many people understand this, which is why we see so many "what speakers should I buy?" threads. They're afraid to trust their own ears, and want reassurance from others who have heard more speakers over a longer time.

--Ethan

In the promotional material on their website for their M2 bookshelf speaker, Axiom links to a review giving a graph of "THD+N", here (http://www.soundstagemagazine.com/measurements/axiom_m2i/), among other specs.

Excellent! If only everyone else did that. Sadly, I see no IMD or off-axis response. :eek:

--Ethan

It's easy to measure a speaker, not so much to measure preference. People sort of know it when they're exposed to it.

Agreed full. There's also educated listeners and uneducated listeners. If you know anyone who thinks their Bose table radio is an excellent hi-fi, you know what I mean.

If I were to go into speaker making, I would aim for accuracy above all else. I understand this means I won't sell as many to newbies who prefer a response skewed toward "pleasant." Such is life. There's no accounting for taste.

--Ethan

It's easy to measure a speaker, not so much to measure preference. People sort of know it when they're exposed to it.

+1, I missed that opinion....so true!

Agreed full. There's also educated listeners and uneducated listeners. If you know anyone who thinks their Bose table radio is an excellent hi-fi, you know what I mean.

If I were to go into speaker making, I would aim for accuracy above all else. I understand this means I won't sell as many to newbies who prefer a response skewed toward "pleasant." Such is life. There's no accounting for taste.

--Ethan


Those have been my points above, education is key and measurements are the tools for that.

DIYers always are building speakers with the flattest curves, lowest distortion, least amount of box coloration. I never get all this 'timbre' talk in the DIY world and that does tell me something about the different groups out there.

As for personal preference. I built speakers using active crossovers because then I have everything I will ever need...if I want accuracy, I have a setting, if I want bright/warm/boom box, etc, I have a setting, if want cymbals, I have a setting :D

The right speaker is only the right speaker for a little while usually. If they listening long enough they will start to realize its imprefect and they will want to upgrade.



The implication then, is that this research will lead to only one "right" speaker that everyone will buy (assuming it's not cost prohibitive) and thus no need to ever upgrade?

IMO, there will never be a consolidation of preference and the "best" speaker.

The problem with best is that there is always tomorrow when today's best is no longer best. The problem with right is that what's right for you may not be right for me. There is nothing wrong with liking a speaker that is voiced some particular way, which seems to be another (unintentional) implication. That said, I am of the accurate reproduction camp, but some like a particular sound regardless of pure accuracy.

Perhaps all that is too philosophical for the gist of the tread?

It's easy to measure a speaker, not so much to measure preference.

According to Floyd Toole, speakers that measure the best also sound the best to both experienced and inexperienced listeners. Thus, it is possible to measure preference and, fortunately, there is a correlation between measurements of performance and measurements of preference. This means that the measured performance can be a predictor of preference if the measurements are those with a high degree of correlation. Unfortunately, manufacturers will likely publish only the measurements which make them look good, as they do now.

Unfortunately, manufacturers will likely publish only the measurements which make them look good, as they do now.

No sh!t. :D

Famous old quote, Either a poster in Einstein's office, or he said it:
"Everything you can count, does not count.
Everything that counts, you can't count."

The implication then, is that this research will lead to only one "right" speaker that everyone will buy (assuming it's not cost prohibitive) and thus no need to ever upgrade?

IMO, there will never be a consolidation of preference and the "best" speaker.

The problem with best is that there is always tomorrow when today's best is no longer best. The problem with right is that what's right for you may not be right for me. There is nothing wrong with liking a speaker that is voiced some particular way, which seems to be another (unintentional) implication. That said, I am of the accurate reproduction camp, but some like a particular sound regardless of pure accuracy.

Perhaps all that is too philosophical for the gist of the tread?

You are right about the subjective "best", we even change our preferences over our lives so its a moving target.

I also posted before that Im okay with someone liking a particular sound as long as they understand what accuracy sounds like they can make any decision they want. This is not about forcing people to buy something, it should be about helping them.

I think you didn't read about the science part ;) and why do people insist its idiotic? What about the thousands of DIYers and speaker builders out there that build speakers based 100% on the numbers? Are they idiotic?

Nope. I'd say it's idiotic (sorry, sivadselim, but that's the word you introduced) to ignore the body of work done on measurements as predictors of speaker preference.

Based on this work, it's a distinct possibility -- even likely -- that sivadeselim's preference, and mine, would be for the more 'accurate' (by NRC-type criteria) loudspeaker...if the comparison was done fairly (WHICH WE TYPICALLY CANNOT DO AT HOME OR IN STORES). And, as happened for some of the Toole/Olive listeners, the 'blind' choice might even represent a *change* in preference, compared to his 'sighted' choice.

So to my mind, playing the odds by setting measured performance as the fundamental criterion, is far from idiotic. It's an extension of the logic for picking 'commodity gear' that isn't even likely to sound *different* (cables, CDPs, and with some further provisos, amps). The 'objectivist' logic there is to buy based on features and price, not sound, because the measurements don't typically indicate there SHOULD be a sonic difference in a fair comparison..in which case preference is actually being driven by nonaudible factors. AIUI, sivadselim adheres to this logic...for those classes of gear.

With loudspeakers, the measurements DO indicate sonic difference is likely between two random sets of loudspeakers, compared fairly. Therefore the sound will be a factor (diluted by biasing confounders) in preference. If we had no data about how the measurable differences in sound are likely to relate to listener preference, then it WOULD be foollish to choose speakers without hearing them. But we DO have data about that.

And btw, I am not talking about manufacturer 'specs' when I write of loudspeaker measurements. I am talking about independent NRC-type evaluation...though some makers do that sort of measurement in-house, AND also do blind listening trials with trained listeners. Would anyone consider it *idiotic* to place value on that?

I'll give another whirl at a post an see if this one gets read. :rolleyes:


If I were to go into speaker making, I would aim for accuracy above all else. I understand this means I won't sell as many to newbies who prefer a response skewed toward "pleasant." Such is life. There's no accounting for taste.

--Ethan

Please define accuracy per your above list. How close do you figure you can actually get to this definition?


DIYers always are building speakers with the flattest curves, lowest distortion, least amount of box coloration. I never get all this 'timbre' talk in the DIY world and that does tell me something about the different groups out there.

It's easy to set targets for "flat" when you only target a very limited set of parameters. Most all of these designs have their own colorations inherent the design choices or executions while looking flat in the particular metrics the designer chose.

The implication then, is that this research will lead to only one "right" speaker that everyone will buy (assuming it's not cost prohibitive) and thus no need to ever upgrade?

IMO, there will never be a consolidation of preference and the "best" speaker.

I fully agree, as there will always be aspects of size, output, cost, etc that make this a practical impossibility, and when you then add the very different acoustic environments we use speakers in, it's nearly impossible to come up with one singular solution. This is still before we add the listener preferences for subjective spaciousness, specificity, etc, etc. You have a little more room for consolidation with a known acoustic environment and multi-channel system, but for 2ch stereo, all bets are off.

Thus, it is possible to measure preference and, fortunately, there is a correlation between measurements of performance and measurements of preference. This means that the measured performance can be a predictor of preference if the measurements are those with a high degree of correlation. Unfortunately, manufacturers will likely publish only the measurements which make them look good, as they do now.

Two very good points here Kal.

First, I would emphasize the point that you can use measurements as effective predictors of likely(not certain) preference and more so of characteristic differences, especially once the listener establishes some correlation with two notably different examples.

You are also correct that more than not, you can expect the preferential measurements, where they aren't often lies, but more so convenient omissions from manufacturers. There are third party labs now available, created for the pro audio industry, which could be very useful for end users, where manufacturers, users, or reviewers can submit their products for testing. Of course there is still some limit to the scope of any set of measurements, and the ultimate hurdle still lies in interpretation, as even the "experts" will have somewhat different takes on what a set of data tells us.

Before we get too far off track, I do want to be clear that simulation and measurements are a HUGE part of my own design efforts. Listening is through the process is mostly to check for offenses possibly missed in the metrics accounted for. Only at the end when the design is functioning as desired do I get into the final subjective refinements as there are an infinite number of different sounding curves within a +/-2dB window.

My point is for posters to understand that at this point we are mostly looking at only a sampling of the full acoustic behavior, and the "rendering" of those details is still done in our mind. I do believe we can do much better, but we're still a long way from any sort of agreement on precisely what matters, what is better/worse and the correlation to subjective impressions.

The implication then, is that this research will lead to only one "right" speaker that everyone will buy (assuming it's not cost prohibitive)

Not 'everyone' -- science is *statistical* -- but the trend is there that listeners, including 'trained' listener, (but perhaps not 'audio professionals' like mixers who listen for minute detail for a living), tend to PREFER the sound of speakers that measure a certain way, and not others ways...IF they are given the chance to compare them fairly, which they NEVER ARE in normal course of life.

THis is assuming that ONLY sound is being considered as an indicator of what's 'right'. Regardless of what they tell themselves, people are rarely ONLY buying for sound. They typically also have price, looks, and other factors in mind -- either consciously, or unconsciously.


IMO, there will never be a consolidation of preference and the "best" speaker.



Well, there the data so are aren't going your way. No one can make a 100% sure call about an individual listener, but a 'consolidation of preference' -- or more accurately, a convergence of preference onto loudspeakers measuring within certain limits in certain parameters -- is exactly what Olive/TOole data points to, for listeners on average.

Before we get too far off track, I do want to be clear that simulation and measurements are a HUGE part of my own design efforts. Listening is through the process is mostly to check for offenses possibly missed in the metrics accounted for. Only at the end when the design is functioning as desired do I get into the final subjective refinements as there are an infinite number of different sounding curves within a +/-2dB window.

Bingo!

My reading of the data is there is not one precise magic number for the currently-defined 'good measurements'. Loudspeakers can still measure significantly differently but still be within a range that is statistically likley to 'sound good'. All 'good' graphs are not exactly alike.


My point is for posters to understand that at this point we are mostly looking at only a sampling of the full acoustic behavior, and the "rendering" of those details is still done in our mind. I do believe we can do much better, but we're still a long way from any sort of agreement on precisely what matters, what is better/worse and the correlation to subjective impressions.

Well devil-detail words like 'precisely' set the bar very high.

I'll give another whirl at a post an see if this one gets read. :rolleyes:




lmao, I did read the first ;)

What part of sound is not measureable?the good part

It's easy to set targets for "flat" when you only target a very limited set of parameters. Most all of these designs have their own colorations inherent the design choices or executions while looking flat in the particular metrics the designer chose.


Yes, they do but we understand through the measurements posted never by go out and listening. I find the people that build speakers probably trust the data more because their data is much more accurate then what is put out there by speaker companies (I agree with many posts on the fact that sadly measurements are not that accurate). I still think we should work towards accurate measurements that can predict speaker performance and therefore help a large majority of the population.


I fully agree, as there will always be aspects of size, output, cost, etc that make this a practical impossibility, and when you then add the very different acoustic environments we use speakers in, it's nearly impossible to come up with one singular solution. This is still before we add the listener preferences for subjective spaciousness, specificity, etc, etc. You have a little more room for consolidation with a known acoustic environment and multi-channel system, but for 2ch stereo, all bets are off.



No need for a signular solution but we can all list our individual needs and go through all the data to match the speakers that will fit our needs the best. Will I find the exact speaker through measurements? That was never my intent but I do want to believe that measurements give me the sound qualities I desire.

I still believe the average person does not have a clue what they should be listening for and therefore seldom picks the right speaker. Down the road they might figure it out and they might now. If we taught them upfront what great measurements look like they might have listened and learned then to pick the better speaker.

I do not think the discussion here is geared towards people like Miles or HiFi or Rutgar...those individuals have spent years already educating themselves on what truely sounds good so they have an "old fashioned" appoarch to speaker selection. Their methods are fine for them but why should they believe that selection method would work for others, how many years did it take them to get it right? Maybe everyone should take years to get it right? Im a big believer in creating new methods so that future audiophiles need not go through all the same mistakes past audiophiles have made.

This does fly in the face of the standard audiophile idea of going into a local audio shop and listening but on the other hand. That type of industry is dying a slow death and at some point the majority of people will only have in house demos to work with. I believe by then we should have measurements in place so that people have a better chance of success and less money spent returning stuff.

the good part

Thanks, I think you have used up your post count in this thread ;)

Excellent! If only everyone else did that. Sadly, I see no IMD or off-axis response. :eek:

The first two charts show response at 0, 15, 30, 45, 60, and 75 degrees off-axis. Or maybe you mean something different? By the way, Axiom gives similar links to NRC measurements for many of their speakers.

Thanks, I think you have used up your post count in this thread ;)OK, fine.

What part of sound is not measureable?no part


(you misspelled measurable, btw)

OK, fine.

no part


(you misspelled measurable, btw)

damn spellcheck ;)

OK, so how about we take another approach at this thread? What measurements does everyone think actually DO predict the quality of sound from a set of speakers? I'll take a first stab - someone tell me if I'm anywhere close....

> Frequency response, on-axis, 1M, 1W
> FR, 15 degrees lateral, 1W, 1M
> FR, 30 degrees lateral, 1W, 1M
> FR, 45 degrees lateral, 1W, 1M
> FR, on-axis, 10W, 1M
> FR, on-axis, 100W, 1M
> Onset of distortion (in dB or input power) vs. frequency [What's a reasonable value for "onset"? 1%?]
> Distortion vs frequency, 1W, 1 M, on-axis
> Electrical impedance vs frequency
> Sensitivity - total
> System resonance (decay time vs. frequency)

Does this get us enough info to begin to understand what the system will sound like?

OK, so how about we take another approach at this thread? What measurements does everyone think actually DO predict the quality of sound from a set of speakers? I'll take a first stab - someone tell me if I'm anywhere close....

> Frequency response, on-axis, 1M, 1W
> FR, 15 degrees lateral, 1W, 1M
> FR, 30 degrees lateral, 1W, 1M
> FR, 45 degrees lateral, 1W, 1M
> FR, on-axis, 10W, 1M
> FR, on-axis, 100W, 1M
> Onset of distortion (in dB or input power) vs. frequency [What's a reasonable value for "onset"? 1%?]
> Distortion vs frequency, 1W, 1 M, on-axis
> Electrical impedance vs frequency
> Sensitivity - total
> System resonance (decay time vs. frequency)

Does this get us enough info to begin to understand what the system will sound like?


Thats a great start, thanks!

The system resonance though is a room dependant isnt it?

Thats a great start, thanks!

The system resonance though is a room dependant isnt it?

Toal resonance would be, but any resonance(s) in the drivers or cabinet could be measured in an anechoic chamber with an impulse.

Toal resonance would be, but any resonance(s) in the drivers or cabinet could be measured in an anechoic chamber with an impulse.

I agree so system resonance is box resonance and driver resonance.

I agree so system resonance is box resonance and driver resonance.

Right - the "speaker" is a system consisting of drivers, crossover, enclosure.

OK, so how about we take another approach at this thread? What measurements does everyone think actually DO predict the quality of sound from a set of speakers? I'll take a first stab - someone tell me if I'm anywhere close....

> Frequency response, on-axis, 1M, 1W
> FR, 15 degrees lateral, 1W, 1M
> FR, 30 degrees lateral, 1W, 1M
> FR, 45 degrees lateral, 1W, 1M
> FR, on-axis, 10W, 1M
> FR, on-axis, 100W, 1M
> Onset of distortion (in dB or input power) vs. frequency [What's a reasonable value for "onset"? 1%?]
> Distortion vs frequency, 1W, 1 M, on-axis
> Electrical impedance vs frequency
> Sensitivity - total
> System resonance (decay time vs. frequency)

Does this get us enough info to begin to understand what the system will sound like?

FR vs. power levels.
Distortion vs. power levels.
Output vs. power levels (i.e., power compression)

...convergence of preference...

Thanks. That is a much better way to put it.

Penngray -

Maybe "flawed" was not the best choice of words, but I do think the notion of selecting speakers off of measurements is idealistic. While I do believe that it is possible to show general trends in speaker preferences based on measurements, as Floyde Tool has shown, it is a complicated matter that too many armchair engineers reduce down to frequency response. FR does certainly contribute to sound of a speaker, but equally if not more important is the distortion spectrum and resonances of the speaker, as Ethan pointed out. You might have a speaker with very linear FR and low distortion, but the spectrum of distortion is of the dissonant high-order type, making the speaker sound less pleasing than one with higher distortion of the 2nd order type. Good luck on finding that data.

Then there is how all of the objective data you've collected combines and interacts. I'd say that it is far more complicated to try to interpret the interactions of the speaker's measured characteristics than it would be to simply listen.

I have been attempting to line-up my preferences for speakers based on measurements for the last 10 years, and the currently available measurements published in Stereophile and Soundstage simply don't provide the whole picture (not enough data on distortion for instance ), and even if they did, it comes back to how do all of the measured characteristics interact?

Let me put it this way, I have owned or listened to many if not all of the best measuring speaker available over the past 10 years (KEF Reference, Revel, Dynaudio, Aerial Acoustics, to name a few) and they all had strengths and weakness that had seemingly little to do with the commonly published measurements. Either that, or I was unable to predict how the small deviations from accurate would play against each other, my room, and music I listen to. Why did the Revels sound somewhat bright and "clinical" and the Aerials sound dull and boxy, when they both have extremely linear on and off axis FR? I'm sure that with the proper study we could find the answers, but we aren't there yet.

One last thing that I think undermines the ability to use measurements as a speaker purchasing tool, and thats recording quality. If you don't listen to only the best recordings, you may seek out speakers that minimize the flaws in the recordings you commonly listen to, or you might even chose speakers that equalize the sound. For instance, if you listen to a lot of rock / pop that is typically pretty bright and compressed, you may prefer a less technically accurate speaker that compliments the style of music you listen to.

Penngray -

Maybe "flawed" was not the best choice of words, but I do think the notion of selecting speakers off of measurements is idealistic. While I do believe that it is possible to show general trends in speaker preferences based on measurements, as Floyde Tool has shown, it is a complicated matter that too many armchair engineers reduce down to frequency response. FR does certainly contribute to sound of a speaker, but equally if not more important is the distortion spectrum and resonances of the speaker, as Ethan pointed out. You might have a speaker with very linear FR and low distortion, but the spectrum of distortion is of the dissonant high-order type, making the speaker sound less pleasing than one with higher distortion of the 2nd order type. Good luck on finding that data.

Then there is how all of the objective data you've collected combines and interacts. I'd say that it is far more complicated to try to interpret the interactions of the speaker's measured characteristics than it would be to simply listen.

I have been attempting to line-up my preferences for speakers based on measurements for the last 10 years, and the currently available measurements published in Stereophile and Soundstage simply don't provide the whole picture (not enough data on distortion for instance ), and even if they did, it comes back to how do all of the measured characteristics interact?

Let me put it this way, I have owned or listened to many if not all of the best measuring speaker available over the past 10 years (KEF Reference, Revel, Dynaudio, Aerial Acoustics, to name a few) and they all had strengths and weakness that had seemingly little to do with the commonly published measurements. Either that, or I was unable to predict how the small deviations from accurate would play against each other, my room, and music I listen to. Why did the Revels sound somewhat bright and "clinical" and the Aerials sound dull and boxy, when they both have extremely linear on and off axis FR? I'm sure that with the proper study we could find the answers, but we aren't there yet.
One last thing that I think undermines the ability to use measurements as a speaker purchasing tool, and thats recording quality. If you don't listen to only the best recordings, you may seek out speakers that minimize the flaws in the recordings you commonly listen to, or you might even chose speakers that equalize the sound. For instance, if you listen to a lot of rock / pop that is typically pretty bright and compressed, you may prefer a less technically accurate speaker that compliments the style of music you listen to.

Definitely Idealistic :D

I also understand your point of view because I have read many, many posts from yourself and others. The experience you have goes beyond what I think Im trying to do here. I guess I should focus this idea on people new to the hobby because they are the ones that have no idea what to even start to look for and you raised other great points about bad recordings, types music, etc.

The part I put in bold is actually what the title was about initially. If we see almost identical meaurements of two speakers but they sound completely different what did we miss? The idea is to find the measurement that tells us why they are different .

One last thing that I think undermines the ability to use measurements as a speaker purchasing tool, and thats recording quality. If you don't listen to only the best recordings, you may seek out speakers that minimize the flaws in the recordings you commonly listen to, or you might even chose speakers that equalize the sound. For instance, if you listen to a lot of rock / pop that is typically pretty bright and compressed, you may prefer a less technically accurate speaker that compliments the style of music you listen to.

Interesting. I actually take the opposite approach. I have a group of recordings that I use for both auditioning and system tuning. While a couple of them are very good recordings, several have distinct flaws; I know that if those flaws are audibly present but not enhanced or pushed to the forefront, the sound is fairly well balanced in that region. [These flaws are things like tuning issues, pick/fret noise and fingersqueaks on guitar, breath pops and sibilance, etc. They are minor and probably irrelevant to most listeners, but the fact that they are intentionally de-emphasized in the mix makes them useful.]

While it is clearly useful to use the type of music you are most fond of when audition a system or component, I would submit that you learn more by using a variety of recordings of differing style and mastering approaches.

Please define accuracy per your above list. How close do you figure you can actually get to this definition?

As close as possible! The point is that my list at least defines what matters. Versus the attitude of some who believe there's no way to define what "good" is.

I'll leave it to speaker designers to see how close they can get.

--Ethan

As close as possible! The point is that my list at least defines what matters. Versus the attitude of some who believe there's no way to define what "good" is.

I'll leave it to speaker designers to see how close they can get.

--Ethan

As close as possible to what? You posted a starting point earlier including:


I'm not a speaker expert, so this list may be incomplete:

* On-axis frequency response.

* Off-axis response, not only side to side but also top to bottom, at every angle from 0 to 360 degrees.

* THD, and also IMD at many frequency combinations to assess all drivers and all crossover components. Distortion should also be measured at several output levels from moderately soft to as loud as the speaker can play just short of damage.

* Ringing, again at enough frequencies to assess all drivers and all crossover components.

* Maximum output level versus frequency.

--Ethan

While I agree some of these are solid starting points and can provide valuable information to digest, I strongly object to the notion that there is some known ideal for all of these measurements.

Most would argue that you need to scrap the THD specs, as we know those have poor correlation to sound quality. Guidelines have been suggested by the papers posted on Earl Gedees's site, but good luck on getting a clear consensus on a numeric metric of this, as it really still needs further research. There is more confidence in the contributing factors now, and some metrics have been proposed, but we're a long way from there being a widely accepted agreement.

We also know that narrow variations in magnitude response, particularly dips, have much lower audibility than broad band trends. We know that power response (total radiated energy into all angles) can be just as significant to subjective balance. Directivity and pattern control vs. frequency have significant effects as well. We have zero consensus on what the target is for power response or off axis behavior and the related pattern vs. frequency.

I guess my point to be made here comes back to appropriately setting expectations. While we can be blissfully content in chasing a couple of easy targets, the reality is that we are so far from an ideal loudspeaker that our subjective perceptions are much more related to the balance and combination of strengths and faults than to how exceptional a speaker is in one particular regard.

I would argue it is dangerous to set expectations of making clear delineations with a limited set of data which is highly inter-dependent with respect to subjective correlation (one measurement can affect the significance of another). We have already seen some of that in the subwoofer realm where a few get fixated on specific aspects leaving them caught on the part rather than the whole which we actually listen to. I do agree more data will be a good thing, but how it is presented and explained with differing viewpoints is just as important as its existence.

The short answer is that speakers are so non-ideal that they are almost completely unmanagable, if sonic transparency is the goal.
I wonder what manageability is. Is it the ability to establish a rational pricing, so that if the test scores for speaker A exceed those for speaker B, we would agree that A should cost more than B, and in general, the tests should give a linear ordering of speaker worth?

As close as possible to what?

As close as possible to a flat response both on and off axis, minimal ringing, zero distortion even at high SPL levels, and so forth.

Most would argue that you need to scrap the THD specs

Most who? Distortion - especially IMD which is always present along with THD - is always audibly damaging if it's loud enough to hear. I'd say anything over 1 percent is audible, though there are many variables.

We also know that narrow variations in magnitude response, particularly dips, have much lower audibility than broad band trends.

IMO that's a myth based on faulty testing. More here:

Audibility of Narrow-Band EQ (http://www.realtraps.com/art_narrow.htm)

one measurement can affect the significance of another

Okay, but we still want the response as flat as possible, and distortion as low as possible. We also want off-axis response as flat as possible, at least for the entire area where people may be listening.

I guess I don't understand your point. How could one argue that we should aim for anything other than the ideal?

--Ethan

This is where I will be a fly on the wall and just learn something from the Experts :D

I hope you guys can keep the discussion going!

It's been done: measured performance of a speaker versus listener rating of the sound quality. Look up the work fo Sean Olive and Floyd Toole, across a few decades of JAES publication. Also look up references to 'Canadian NRC' in brochures by Paradigm, JBL, and some other loudspeaker manufacturers.

Flat response with a slight downward tilt starting around 10k? Yeah, I know.

It's easy to predict that non-offensive sound will not offend I suppose.

Flat response with a slight downward tilt starting around 10k? Yeah, I know.

er, no, I guess you don't.



It's easy to predict that non-offensive sound will not offend I suppose.

Shouldn't a loudspeaker aim to be 'non-offensive'?

The 'offensiveness' of a sound should be inherent in the recording, not an artifact of the playback gear.

er, no, I guess you don't.

Yes, I do. It's generally referred to as "anechoically flat" as opposed to the "flat" one can achieve with an RTA, which is almost universally referred to as objectionably bright.

The numbers can shifty a little here and there, but generally, blind tests tend to show a flat response without a ton of brightness, with a smooth low-end. Someone mentioned Paradigm.

The bottom line in this discussion is if/when you measure a speaker's specs and determine that it is either "right", "wrong", "better", "worse", etc. someone else will not agree with you. The reason is because everyone's hearing is different. When you hear a musical instrument in a live unamplified setting you develop an opinion on how this instrument is suppose to sound. This in turn will affect how "accurate" a speaker sounds based on what you heard when this instrument is reproduced by a speaker vs. the live version you heard.

This leads to the fundamental reason as to why measurements no matter how complex or interpreted can not be used to pick speakers to meet your sound requirements. If two people are looking for speakers with the same sound and choose based on what the measurements tell them, they will more than likely still hear the speaker's sound reproduction differently. The reason is a slew of different factors.

The shape of a person's outer ear, ear canal, and inner ear are different from anyone else's. This alone will change the sound heard by two different people no matter what the measurements say. A lifetime of hearing sounds and interpreting them affects how and what we hear. The ears ability to pick up sounds and amplify them to the brain over a frequency range is different for everyone.

The biggest factor is what a person finds pleasing to their ear. This is no different than what a person likes to smell or taste. No one can say that because certain foods have certain ingredients in specific amounts that it is right and all people will either like it or dislike it. Hearing sounds is no different. Just because the measurements say this speaker does everything right does not mean that it will do everything right for every person or even 2 people.

What is accepted as "right" may be all wrong to that person even when they are considered "educated" in high quality sound. Because every person hears differently, what their brain interprets as "right" may sound all wrong to the next person.

There is no way to use measurements to say this speaker is "accurate" or "sounds right" or anything else because of these facts. Whether you believe this or not doesn't change the fact that it is true. Every single person in the world hears differently. If there was a formula to make the perfect speaker then someone would have either found it or come very close. But it is not possible to compensate for how everyone's brain interprets the signals from the ear, let alone how the sounds are affected by the shape and movement of the ear.

Measurements can get you in the ball park or maybe in the dugout, but that is all. This is why there are so many speaker brands and designs available. Because no two people will hear the same thing.

Yes, I do. It's generally referred to as "anechoically flat" as opposed to the "flat" one can achieve with an RTA, which is almost universally referred to as objectionably bright.

The numbers can shifty a little here and there, but generally, blind tests tend to show a flat response without a ton of brightness, with a smooth low-end. Someone mentioned Paradigm.
Perhaps the reason it's bright has to do with problem that exist in the time domain?

The bottom line in this discussion is if/when you measure a speaker's specs and determine that it is either "right", "wrong", "better", "worse", etc. someone else will not agree with you. The reason is because everyone's hearing is different. When you hear a musical instrument in a live unamplified setting you develop an opinion on how this instrument is suppose to sound. This in turn will affect how "accurate" a speaker sounds based on what you heard when this instrument is reproduced by a speaker vs. the live version you heard.

This leads to the fundamental reason as to why measurements no matter how complex or interpreted can not be used to pick speakers to meet your sound requirements. If two people are looking for speakers with the same sound and choose based on what the measurements tell them, they will more than likely still hear the speaker's sound reproduction differently. The reason is a slew of different factors.

The shape of a person's outer ear, ear canal, and inner ear are different from anyone else's. This alone will change the sound heard by two different people no matter what the measurements say. A lifetime of hearing sounds and interpreting them affects how and what we hear. The ears ability to pick up sounds and amplify them to the brain over a frequency range is different for everyone.

The biggest factor is what a person finds pleasing to their ear. This is no different than what a person likes to smell or taste. No one can say that because certain foods have certain ingredients in specific amounts that it is right and all people will either like it or dislike it. Hearing sounds is no different. Just because the measurements say this speaker does everything right does not mean that it will do everything right for every person or even 2 people.

What is accepted as "right" may be all wrong to that person even when they are considered "educated" in high quality sound. Because every person hears differently, what their brain interprets as "right" may sound all wrong to the next person.

There is no way to use measurements to say this speaker is "accurate" or "sounds right" or anything else because of these facts. Whether you believe this or not doesn't change the fact that it is true. Every single person in the world hears differently. If there was a formula to make the perfect speaker then someone would have either found it or come very close. But it is not possible to compensate for how everyone's brain interprets the signals from the ear, let alone how the sounds are affected by the shape and movement of the ear.

Measurements can get you in the ball park or maybe in the dugout, but that is all. This is why there are so many speaker brands and designs available. Because no two people will hear the same thing.


I still think people like to be subjective just because they can be....they hate being the same as others and they will argue to no end that a minor half percent change in sound is night and day to them just because they can argue it.

Besides this thread is not about building perfect speakers because we can build scientificially measured accurate speakers already for those who want that.

I tend to move away from any subjectivity and I let science decide what a good speaker is for me. That is why I enjoy DIY so much more then even going out and listening to any speaker. In the end if I can get accurate measurements out of my speaker Im going to enjoy the crap out of it. I have no need to be fussy in what I want, I just want to have something I know is absolutely good and if I want to have a slightly different sound I would tweak in room.

Besides our personal tastes. The point from the start is to build a set of measurements that will guide people new to the hobby that have not created specific preferences in their audio habits. Its not about helping anyone that has formed an opinion because years of experience.

Or more to the point only new people can be helped. ;)

Measurements can still dictate the sonic nature of speakers. I do not care about the differences in ears either because if a speaker does not color sound its going to sound the great for me and great for anyone else because its not altering the sound for anyone. The only question here is there something that can not be measured? Some people say yes but Im not sold on those points.

How about this idea....

If people truely want to distort the FR then that is cool, they can use EQing to do it but the speakers should be built perfectly flat, with low distortion and almost zero box coloration. Then in room anyone can tweak the speakers to what they want them to sound like.

Perhaps the reason it's bright has to do with problem that exist in the time domain?

Creating a speaker that fits what 95% of the population likes isn't tough. It's the esoteric brands, driven by designers with some sort of "vision" of a specific type of sound that things get wonky.

Creating a speaker that fits what 95% of the population likes isn't tough. It's the esoteric brands, driven by designers with some sort of "vision" of a specific type of sound that things get wonky.

Very true so why does 95% of the population pick from 100s of different brands?

Its probably as simple as saying people do not like wearing the same clothes as others.

This is why I believe there is great debate about the idea that measurements can predict the performance of most speakers. If people actually agree with that notion they are going against their own subjective well being in some way.

Creating a speaker that fits what 95% of the population likes isn't tough. It's the esoteric brands, driven by designers with some sort of "vision" of a specific type of sound that things get wonky.A person once said, "Any idiot can design a loudspeaker and unfortunately many do."

Yes, I do. It's generally referred to as "anechoically flat" as opposed to the "flat" one can achieve with an RTA, which is almost universally referred to as objectionably bright.

1) Most rooms don't have effective treatments.
2) Any harmonic distortion increases the apparent/perceived brightness.

This doesn't mean that a non-flat speaker is "better" - just that other issues combine to cause problems in the higher frequencies.

The 'offensiveness' of a sound should be inherent in the recording, not an artifact of the playback gear.

Excellent point. Give me a speaker that's as accurate as possible, so I can hear what the mix engineer heard when he or she worked on the project.

1) Most rooms don't have effective treatments.
2) Any harmonic distortion increases the apparent/perceived brightness. This doesn't mean that a non-flat speaker is "better" - just that other issues combine to cause problems in the higher frequencies.

Exactly.

--Ethan

1) Most rooms don't have effective treatments.
2) Any harmonic distortion increases the apparent/perceived brightness.

This doesn't mean that a non-flat speaker is "better" - just that other issues combine to cause problems in the higher frequencies.

Oh I agree totally, I just said that in many DBTs they were "preferred."

Yes, I do. It's generally referred to as "anechoically flat" as opposed to the "flat" one can achieve with an RTA, which is almost universally referred to as objectionably bright.

No, you don't. You're thinking of room curve and why it is employed, whereas the Toole/Olive/NRC measurement set involves more than just on-axis FR.

The bottom line in this discussion is if/when you measure a speaker's specs and determine that it is either "right", "wrong", "better", "worse", etc. someone else will not agree with you.


OK, one more time...

THe bottom line is actually that there will be *someone* who will disagree...but *chance are* that most listeners won't...*IF* you, and they, were actually able to compare loudspeakers fairly.

This was an important finding.

The shape of a person's outer ear, ear canal, and inner ear are different from anyone else's. This alone will change the sound heard by two different people no matter what the measurements say. A lifetime of hearing sounds and interpreting them affects how and what we hear. The ears ability to pick up sounds and amplify them to the brain over a frequency range is different for everyone. Since this is a fixed parameter for any given individual, the difference between real and reproduced is the same for everyone whose hearing is within the 'normal' range. In other words, imho, this is not a factor.

The biggest factor is what a person finds pleasing to their ear. This is no different than what a person likes to smell or taste. No one can say that because certain foods have certain ingredients in specific amounts that it is right and all people will either like it or dislike it. Hearing sounds is no different. Just because the measurements say this speaker does everything right does not mean that it will do everything right for every person or even 2 people. Right. In matters of taste, accuracy becomes relatively unimportant.

There is no way to use measurements to say this speaker is "accurate" or "sounds right" or anything else because of these facts. Whether you believe this or not doesn't change the fact that it is true.I disagree. There is a HUGE distinction between "accurate" and "sounds right" as you, yourself, have implied above.

This is why I believe there is great debate about the idea that measurements can predict the performance of most speakers. If people actually agree with that notion they are going against their own subjective well being in some way.


That's likely because people aren't typically evaluating ONLY 'performance' (audible performance) in their 'auditions' -- whether they realize it or not.

When they ARE put in a situation where they are judging a loudspeaker purely on 'performance' (audible) -- then things get very interesting....

That's likely because people aren't typically evaluating ONLY 'performance' (audible performance) in their 'auditions' -- whether they realize it or not.

When they ARE put in a situation where they are judging a loudspeaker purely on 'performance' (audible) -- then things get very interesting....

The circlejerk element to this idea though is that even blindfolded, there are still three other senses to contend with (other than hearing) that will affect the perception of the experience.

If the only way we can properly test is to totally deprive peoples' senses so they can only focus on the sound, then it's totally pointless to even try.

The circlejerk element to this idea though is that even blindfolded, there are still three other senses to contend with (other than hearing) that will affect the perception of the experience.

The jerk aspect to this is to equate 'double-blinding' with 'blindfolding'.

And yes, the fact that information from other senses 'affect the perception of the experience' of sound is *THE POINT* of blind listening tests. Subjectivists would do well to acknowledge these biasing effects, but they seem very reluctant to do so. Conclusions drawn about *performance of the gear* from biased 'perceptions of the experience' can be off...from a little bit off, to *laughably wrong*...as in, while you reported that A sounded much better than B, you were actually listening to the same thing twice.

(IOW: It wasn't the gear...it was *you*.)


If the only way we can properly test is to totally deprive peoples' senses so they can only focus on the sound, then it's totally pointless to even try.

Wrong on both counts. So, your grasp of sensory testing derives from...where, exactly? You claim to have read enough subjectvist vs objectivist threads to drive you to advocate censorship...but it doesn't look like you've actually understood what was written.

Thus, it is possible to measure preference and, fortunately, there is a correlation between measurements of performance and measurements of preference.


I guess what you're saying is that there is such a thing as general preference. We also know that there is such a thing as individual preference. Individual preference is well, individual so therefore it is difficult or impossible to predict.

However, we may be learning for speakers what we already knew for amplifiers, and that there is a general preference for speakers whose measurements conform to a certain form.



This means that the measured performance can be a predictor of preference if the measurements are those with a high degree of correlation.


In the end the only kind of preference that speaker manufacturers are really interested in is the kind of perference that correlates with people spending money to obtain their speakers.


Unfortunately, manufacturers will likely publish only the measurements which make them look good, as they do now.

What we are hoping for is a change from the current pattern of manufacturers trying to sell speakers based on incomplete data that is chosen to make their speakers look good on paper, into a pattern of speakers that tend to sound good to the largest possible group of potential purchasers.

There is no way to use measurements to say this speaker is "accurate" or "sounds right" or anything else because of these facts. Whether you believe this or not doesn't change the fact that it is true. Every single person in the world hears differently. If there was a formula to make the perfect speaker then someone would have either found it or come very close. But it is not possible to compensate for how everyone's brain interprets the signals from the ear, let alone how the sounds are affected by the shape and movement of the ear.

Measurements can get you in the ball park or maybe in the dugout, but that is all. This is why there are so many speaker brands and designs available. Because no two people will hear the same thing.

Since this is a fixed parameter for any given individual, the difference between real and reproduced is the same for everyone whose hearing is within the 'normal' range. In other words, imho, this is not a factor.

...

I disagree. There is a HUGE distinction between "accurate" and "sounds right" as you, yourself, have implied above.

Great response Kal, and much more direct than my explanations tend to be. :rolleyes:

I have no question we all process the sounds that arrive at our ears a little differently. There is plenty of evidence supporting this. The problem is perspective. So long as we think of speakers being even close to ideal, rather than the compromised reproduction devices they are, we get stuck on the smaller picture.

If we go to an outdoor concert and hear a trumpet and a flute, two people with rather different hearing qualities standing next to each other will both identify them as a trumpet and flute. If we then hear a french horn and a piccolo, both listeners will identify them as similar, but different from the trumpet and flute. The important part to notice here is that both listeners were presented with the SAME acoustic event!

They may judge the sounds being produced as more or less desirable, but if the same trumpet and flute played the same piece again, they could both agree it was the same as the first time it was played.

The important realization here is that we all are "sampling" from the same total sound arriving at our ears. There are both physical properties which define what we hear as well as those related to mental focus and processing of what is received. If we stick to this reference of an original event, and we then try and replicate this event, it would make sense that one listener might perceive a deficiency in reproduction that another would not or wouldn't as easily.

We still have a lacking reproduction, but for one listener the information they sample from the event is almost entirely reproduced. A different limitation could flip the observation. Consider the very simple diagram below.
http://www.seaton-sound.com/mark/event-v-listeners-sm.jpg

My own experience suggests that the more faults which can be removed, the wider the range of listeners who are likely to enjoy the result, and the wider range of uses a speaker will be suitable for. Adding to the above diagram, you can imagine the bubble of what our loudspeakers reproduce being a circle within the Acoustic Event which largely, but not completely, includes both listener circles, leaving deficiencies that both can hear as well as a few that either are more sensitive to than the other.

The jerk aspect to this is to equate 'double-blinding' with 'blindfolding'.

Thanks, I understand the difference. Try to stop being such a robot.

And yes, the fact that information from other senses 'affect the perception of the experience' of sound is *THE POINT* of blind listening tests. Subjectivists would do well to acknowledge these biasing effects, but they seem very reluctant to do so.

I guess the issue always comes back to the fact that people tend to not listen in controlled, scientific environments, and that mood and input from other senses will always affect perception. Therefore, attempting to "prove" something in a scientific environment tends to only be relevant in that setting. Once removed, biases return and it's back to normal. I don't think (and never have thought) that it's useful to even run these tests, since preferences and situational perceptions will typically dictate anyway.

Wrong on both counts. So, your grasp of sensory testing derives from...where, exactly? You claim to have read enough subjectvist vs objectivist threads to drive you to advocate censorship...but it doesn't look like you've actually understood what was written.

I understand that the threads bounce back and forth, never really offering anything new, except another opportunity to bash someone.

Again, if the sun is shining and the birds are singing, it doesn't mean we're having a good day. Very often, certain data points are not enough to determine anything outside of those data points.

Thanks, I understand the difference. Try to stop being such a robot.


Try to demonstrate that you understand something, rather than just *claiming to*. So far, you're failing. But I've noticed this happen to you on a few threads now, where other 'robots' have called you out on your misstatements.


I guess the issue always comes back to the fact that people tend to not listen in controlled, scientific environments, and that mood and input from other senses will always affect perception. Therefore, attempting to "prove" something in a scientific environment tends to only be relevant in that setting. Once removed, biases return and it's back to normal. I don't think (and never have thought) that it's useful to even run these tests, since preferences and situational perceptions will typically dictate anyway.

So what part of what I wrote -- 'subjectivists would do well to acknowledge these biases' -- does this little rambling reply of yours actually pertain to?

Btw, "proving" something in a 'scientific setting' is actually how knowledge, you know, advances beyond superstition, speculation, 'folk wisdom', and notions.


I understand that the threads bounce back and forth, never really offering anything new, except another opportunity to bash someone.

Thanks for re-iterating the limits of your understanding. And for affirming that 'new to you' is all that matters.


Again, if the sun is shining and the birds are singing, it doesn't mean we're having a good day. Very often, certain data points are not enough to determine anything outside of those data points.

And how do you know which data points fit this description? Instinct?

As close as possible to a flat response both on and off axis, minimal ringing, zero distortion even at high SPL levels, and so forth.

That "and so forth" part is where many head off on chases that leave them compromising the earlier factors you mentioned. :rolleyes:

I'll make the assumption you mean flat anechoic measurement on axis. I can send a speaker out and get full spherical measurements (not just limited horizontal & vertical polars) above 100Hz at 10, 5, 2, 1 or even 0.5 degree increments defining all angles of radiation for the speaker.

So what should this look like in the ideal? You seem to be suggesting that a perfectly omni loudspeaker would be preferred with equal response in every direction.


Most who? Distortion - especially IMD which is always present along with THD - is always audibly damaging if it's loud enough to hear. I'd say anything over 1 percent is audible, though there are many variables.

Many would disagree with you. Earl Gedees and Lidia Lee (GedLee.com) have done a great deal of work on this which takes significant steps forward from observations made 50 years ago and later given much thought in the development of modern compression techniques (regardless of their successes or failures). Please read through some of what has been posted here at the GedLee website (http://www.gedlee.com/distortion_perception.htm).

The short summary is that while distortion can be very audible, THD is a very unreliable metric. Higher order harmonics are audible at dramatically lower levels than say 2nd & 3rd order harmonics, which happen to usually be the most dominant in THD measurements. :rolleyes: I'm not saying high distortion is good, but the examination of "what matters" would be discarding some good sounding options if just looking at THD.

I wouldn't call THD worthless, as it can be used as a first pass to see how fast and at what point in increasing drive level a speaker or subwoofer increases in distortion. A gradual increase in distortion tends to be a different effect than a sharp increase, where further detail will show the nature or cause of either.


IMO that's a myth based on faulty testing. More here:

Audibility of Narrow-Band EQ (http://www.realtraps.com/art_narrow.htm)

OK, I'll give you that was a weak example, and you have some great examples of cases where it does matter (and we'd expect it to). At the same time, I'd much rather have a recession in a loudspeaker's response at 5-7kHz than an equal magnitude peak. Again, you just pointed out the problems inherent to interpreting any such data.

The reality is that we can use fancy DSP to force good measurements in a handful of the metrics you've noted to make things look great, but there can often be little, if any improvement to the subjective sound with such efforts vs. a little more ragged response and dealing with other factors which matter more in the big picture.


Okay, but we still want the response as flat as possible, and distortion as low as possible. We also want off-axis response as flat as possible, at least for the entire area where people may be listening.

I would clarify that we want these aspects as flat/low/good as possible, while not overly compromising other very audible contributors.


I guess I don't understand your point. How could one argue that we should aim for anything other than the ideal?

--Ethan

My point is that "flat" gets a little fuzzy when you start looking at real applications and uses of speakers, let alone the conditions under which you decide to measure the speakers. My own experience and plenty of published studies suggest that the off axis behavior on the whole, not just the forward 60-90 deg, has a huge impact on the sound in real rooms. If you add in a desire for significant dynamic headroom and linearity, maybe even efficiency, a more holistic or systemic solution is likely to make a balance of compromises vs excellence in just one or two aspects.

While I'm sure I'm not the only one who would agree with my approach, there will be plenty of others who have different ideas of what makes for a best compromise, and others who have a narrow enough view that they believe their efforts to already be ideal.

Try to demonstrate that you understand something, rather than just *claiming to*. So far, you're failing. But I've noticed this happen to you on a few threads now, where other 'robots' have called you out on your misstatements.

It's probably a matter of you and those others putting way too much stock into a thread on the internet. Social failure ftl.

So what part of what I wrote -- 'subjectivists would do well to acknowledge these biases' -- does this little rambling reply of yours actually pertain to?

Btw, "proving" something in a 'scientific setting' is actually how knowledge, you know, advances beyond superstition, speculation, 'folk wisdom', and notions.

OK, thanks for the obvious:rolleyes:

Subjectivists probably enjoy listening to the music, not the systems, which I wonder about with the objective crowd, always turning a hobby into a lab experiment. More to your question's point though, the issue is that more than likely, consumers will not magically follow someone else's experiences as a guide when purchasing themselves. So really, whatever findings are involved are a study in wasting oxygen.

Thanks for re-iterating the limits of your understanding. And for affirming that 'new to you' is all that matters.

I think responses like this are perfect examples of what I mean when I say that nerdrage from people like you is the real point of your participation on forums like this. It's nice to have your opinions finally heard, isn't it?

And how do you know which data points fit this description? Instinct?

Do I have to know specifically which ones out of your random experiments are valid to understand that many are not?

I wonder how your response to my statements in your TYPICAL hostile manner is in any way helpful to this (or any) thread? All you do is look petty, insecure, and immature. Again.

The reality is that we can use fancy DSP to force good measurements in a handful of the metrics you've noted to make things look great, but there can often be little, if any improvement to the subjective sound with such efforts vs. a little more ragged response and dealing with other factors which matter more in the big picture.



This coincides with my personal experience. I owned a set of speakers from a well-known French speaker company (Focal Electra 1037 Be) that sounded harsh and unnatural in the upper midrange. Using Audyssey Pro EQ to correct the in-room response had no effect on reducing the "harshness" I heard in that speaker. Likewise, there was nothing in the uncorrected FR measurements that would point to a problem in the upper mids. In fact there was dip in the in-room FR between 1KHz and 4KHz, so if anything this speaker *should have* sounded somewhat relaxed if we go off of the common knowledge that FR is one of the key contributors to sound quality.

Based on that experience, it seems clear to me that there was some form of distortion present in that speaker that did not show up on the graphs published in Stereophile for that speaker, nor in my own measurements (which consisted only of FR and spectral decay).

Well, unless I've overlooked something. Take a look for yourself at the measurements of this speaker in the link below, and let me know if you could have predicted problems in upper midrange from them.

http://www.stereophile.com/floorloudspeakers/707focal/index4.html

Well the relative peak centered at 5khz would be a possible indicater of harshness and sibilence. Not necessarily but worth a look... the fact that you find an upper harshness to these leans that way..

also the measument is smoothed... and averaged over a 20 measurment locations... the offense could be much worse than noted... however I'm sure most would be quite offended if they saw true measurements of their beloved speakers.
:rolleyes:
This coincides with my personal experience. I owned a set of speakers from a well-known French speaker company (Focal Electra 1037 Be) that sounded harsh and unnatural in the upper midrange. Using Audyssey Pro EQ to correct the in-room response had no effect on reducing the "harshness" I heard in that speaker. Likewise, there was nothing in the uncorrected FR measurements that would point to a problem in the upper mids. In fact there was dip in the in-room FR between 1KHz and 4KHz, so if anything this speaker *should have* sounded somewhat relaxed if we go off of the common knowledge that FR is one of the key contributors to sound quality.

Based on that experience, it seems clear to me that there was some form of distortion present in that speaker that did not show up on the graphs published in Stereophile for that speaker, nor in my own measurements (which consisted only of FR and spectral decay).

Well, unless I've overlooked something. Take a look for yourself at the measurements of this speaker in the link below, and let me know if you could have predicted problems in upper midrange from them.

http://www.stereophile.com/floorloudspeakers/707focal/index4.html

You sure those weren't voiced by a French waiter?

Well the relative peak centered at 5khz would be a possible indicater of harshness and sibilence. Not necessarily but worth a look... the fact that you find an upper harshness to these leans that way..



Certainly a possibility, but it seems to me that if the perceived harshness were the result of a peak in the FR at 5KHz, that the harshness would go away if that peak were flattened. As I mentioned, I used Audyssey EQ to flatten the reponse, and the harshness remained.

I'll make the assumption you mean flat anechoic measurement on axis. I can send a speaker out and get full spherical measurements (not just limited horizontal & vertical polars) above 100Hz at 10, 5, 2, 1 or even 0.5 degree increments defining all angles of radiation for the speaker. So what should this look like in the ideal? You seem to be suggesting that a perfectly omni loudspeaker would be preferred with equal response in every direction.

Good point. I don't think a speaker for most home uses should be omni because the rear radiation just makes extra reflections off the front wall that have to be removed with absorption. So let's say the response should be flat over as wide an angle as needed to reach all listeners in the room.

while distortion can be very audible, THD is a very unreliable metric.

But I said "especially IMD" because IMD is always a problem since it contains non-musical components.

Higher order harmonics are audible at dramatically lower levels than say 2nd & 3rd order harmonics, which happen to usually be the most dominant in THD measurements.

I believe that's wrong. A square wave is the most extreme example of clipping and distortion one could fine, and the harmonics roll off at a predictable rate with higher harmonics always being softer than those preceding. Yes, there are some "buzzy" types of distortion you can have with power amplifiers and other electronics. But I'd expect speaker drivers to roll off as you go higher.

I'm not saying high distortion is good, but the examination of "what matters" would be discarding some good sounding options if just looking at THD.

Again, whenever you have THD you also have IMD. Like peas in a pod. That's just the way it works. At least for normal distortion mechanisms.

Good discussion Mark!

--Ethan

there was nothing in the uncorrected FR measurements that would point to a problem in the upper mids. In fact there was dip in the in-room FR between 1KHz and 4KHz, so if anything this speaker *should have* sounded somewhat relaxed

Harshness is often caused by early reflections in the room. Do you have any absorption at those places? Also, there are dips and there are dips. The Q of the dip matters as much as its magnitude. Did you actually measure the speakers in your room? Harshness doesn't have to be at 3-4 KHz, and assessing stuff like this by ear can be difficult.

--Ethan

Subjectivists probably enjoy listening to the music, not the systems, which I wonder about with the objective crowd, always turning a hobby into a lab experiment.

Hmmm, it really seems the other way around, except you could then delete the word "lab."

No kidding Bob. I've been to hi-fi club meetings and all they do is sit around all day listening to this amp versus that amp, this cable versus that cable, and so forth, including before and after warming up their solid state gear. :D

Me, I rarely buy or swap out gear because I'm totally satisfied with both of my systems. So I spend all of my time listening to music, or recording and mixing music. Other than a new Dell computer two years ago, I haven't bought anything new for my studio in many years.

--Ethan

That "and so forth" part is where many head off on chases that leave them compromising the earlier factors you mentioned. :rolleyes:

I'll make the assumption you mean flat anechoic measurement on axis.
.

(snip).
.
.
.


My point is that "flat" gets a little fuzzy when you start looking at real applications and uses of speakers, let alone the conditions under which you decide to measure the speakers. My own experience and plenty of published studies suggest that the off axis behavior on the whole, not just the forward 60-90 deg, has a huge impact on the sound in real rooms.



I would presume that *anyone* on this thread who is talking about what measurements are important in predicting consumer satisfaction with loudspeaker sound, would definitely include off-axis response...because its role is one of the signal findings of Toole et al.s work.

It's probably a matter of you and those others putting way too much stock into a thread on the internet. Social failure ftl.



OK, thanks for the obvious:rolleyes:

Subjectivists probably enjoy listening to the music, not the systems,



LOL. Stereophile, TAS and its ilk are staunchly 'subjectivist' yet the focus of the 'high end' is very strongly on *gear*. D'you think the people who are switching out amps and cables every year or two are really *objectivists*?



which I wonder about with the objective crowd, always turning a hobby into a lab experiment.

You patently don't know squat about 'the objectivist crowd'. Btw, Einstein was a musician, too.


More to your question's point though, the issue is that more than likely, consumers will not magically follow someone else's experiences as a guide when purchasing themselves. So really, whatever findings are involved are a study in wasting oxygen.


...and your ramblings *still* haven't addressed my point about subjectivists not acknowledging the biases that can lead them to being *ludicrously wrong* about audio generally...and the claims for their own 'systems' specifically.


I think responses like this are perfect examples of what I mean when I say that nerdrage from people like you is the real point of your participation on forums like this. It's nice to have your opinions finally heard, isn't it?


Add 'sarcasm vs rage' to the list of things you're confused about.



Do I have to know specifically which ones out of your random experiments are valid to understand that many are not?


So, instinct, then.

I wonder how your response to my statements in your TYPICAL hostile manner is in any way helpful to this (or any) thread? All you do is look petty, insecure, and immature. Again.

:p

I would presume that *anyone* on this thread who is talking about what measurements are important in predicting consumer satisfaction with loudspeaker sound, would definitely include off-axis response...because its role is one of the signal findings of Toole et al.s work.
180 degrees off-axis (meaning away from forward)? I wouldn't have included that. This was news to me.

Harshness is often caused by early reflections in the room. Do you have any absorption at those places? Also, there are dips and there are dips. The Q of the dip matters as much as its magnitude. Did you actually measure the speakers in your room? Harshness doesn't have to be at 3-4 KHz, and assessing stuff like this by ear can be difficult.

--Ethan

Another possibility, but I don't think that was the case with these speakers. I think that it was something unrelated to FR, be that on or off-axis.

Yes, I measured the speakers in my room using REW. For the most part, my in-room measurements of the Focal's looked like the in-room measurements published in Stereophile for Michael Fremer's room, though he has more exessive bass boost than I do. From about 200Hz on up, my measurements look like this:

http://www.stereophile.com/images/archivesart/707Focfig6.jpg

While I do have room treatments (see the link in my sig), I do not have absorption at the first reflection points. However, knowing a little something about the effects of early reflections, to further inverstiagte the matter of harshness with the Focal's, I listened to them in the near-field (about three feet from the speaker) to reduce the effects of early reflections. The harshness remained.

I later hired an acoustician to evaluate my room and to check my sanity with the sound I was hearing, and he concluded that the problem in the upper mids was a character of the speaker, not the room.

Take a look for yourself at the measurements of this speaker in the link below, and let me know if you could have predicted problems in upper midrange from them.

http://www.stereophile.com/floorloudspeakers/707focal/index4.html

How do you define Upper midrange? That speaker seems to have some weird squiggles from about 3khz - 10khhz on axis. 3khz could be called upper midrange. It's off horizontal axis response shows the typical uneveness around 3khz (my guess is the midrange tweeter crossover is somewhere around there. I didn't read carefully, because I expect his post to be ignored in the background babble). It could be that those wiggles, and off axis response caused problems to your ears in your room.

So what did you end up with for speakers? Are comparable measurements available for them? Maybe by comparing the measurements to your preference there is something you can point to as something you like to see in measurements that corresponds to a speaker you enjoy.

The problem is I think not everybody is sensitive to the same things.
Maybe measurements are not showing what those things are.

Maybe it is just easier and more efficient to listen to speakers rather than trying to spend time comparing measurements with your preferences. Sure, use measurements as a guide - but until somebody can identify what measurement, and what squiggles, are relevant to me I'll stick to using my ears. Yes they can be wrong, but ALL speakers are wrong, so I just pick ones that are the least wrong for the money I can afford to spend. I can't tell least wrong just my measurements - perhaps I need more experience comparing observations with measurements, but I'd rather spend that time listening to music on speakers I like well enough.

While I do have room treatments (see the link in my sig), I do not have absorption at the first reflection points. However, knowing a little something about the effects of early reflections, to further inverstiagte the matter of harshness with the Focal's, I listened to them in the near-field (about three feet from the speaker) to reduce the effects of early reflections. The harshness remained.

I later hired an acoustician to evaluate my room and to check my sanity with the sound I was hearing, and he concluded that the problem in the upper mids was a character of the speaker, not the room.

Not picking on you, just happened to respond to two of your posts
in a row.

I think if you are not absorbing first reflections you are asking for problems from most speakers. Sitting nearfield does not remove those reflections. Sure it puts you closer to the speaker, but it also puts you closer to the first reflections. I haven't worked out the geometry, I'll leave that to an interested reader.

You are right - there is something about that speaker that did not appeal to you. The question is what? Why? How? Was it the measurements? Was it your room? Was it your particular preferences? Likely nobody will know. . .

How do you define Upper midrange?
I like the following, because it's so easy to remember (don't know if it's standard): starting at 20Hz, the first 3 octaves are bass (20-40 low bass, 40-80 mid bass, 80-160 high bass), the next 3 are midrange, and the next 3 are treble. Then "upper midrange" would be 640-1280. (The referenced graph shows a phase peak here.)

How do you define Upper midrange? That speaker seems to have some weird squiggles from about 3khz - 10khhz on axis. 3khz could be called upper midrange. It's off horizontal axis response shows the typical uneveness around 3khz (my guess is the midrange tweeter crossover is somewhere around there. I didn't read carefully, because I expect his post to be ignored in the background babble). It could be that those wiggles, and off axis response caused problems to your ears in your room.

So what did you end up with for speakers? Are comparable measurements available for them? Maybe by comparing the measurements to your preference there is something you can point to as something you like to see in measurements that corresponds to a speaker you enjoy.

The problem is I think not everybody is sensitive to the same things.
Maybe measurements are not showing what those things are.

Maybe it is just easier and more efficient to listen to speakers rather than trying to spend time comparing measurements with your preferences. Sure, use measurements as a guide - but until somebody can identify what measurement, and what squiggles, are relevant to me I'll stick to using my ears. Yes they can be wrong, but ALL speakers are wrong, so I just pick ones that are the least wrong for the money I can afford to spend. I can't tell least wrong just my measurements - perhaps I need more experience comparing observations with measurements, but I'd rather spend that time listening to music on speakers I like well enough.

I think that I have pretty well exhausted the possibility that the problem I heard with those speakers was due to frequency response, especially considering that I hired an acoustician to evaluate my room and those speakers in that room, and because even when I EQ'd those speakers, the problem remained. The acoustician, who has many more years of applied experience than I concluded that the problem in the upper mids was a character of the speaker, not the room. In fact, he told me during a conversation at my home that he has measured speakers that look rolled off, but sound bright, and vice versa. My point to bringing this up is that it seems like most of the focus in this forum is that the character of a speaker's sound lies primarily in the frequency response of the speaker, when it has been my recent experience that there is something else at play. I'm guessing that this problem would show up with some sort of detailed distortion analysis, but this sort of thing is not readily available, and I believe such measurements require an anechoic environment.

In the end, I have found, as you state in your last paragraph, that it is easier and more reliable to listen and decide rather than trying to decipher graphs, which may not even include all of the pertinent info.

I like the following, because it's so easy to remember (don't know if it's standard): starting at 20Hz, the first 3 octaves are bass (20-40 low bass, 40-80 mid bass, 80-160 high bass), the next 3 are midrange, and the next 3 are treble. Then "upper midrange" would be 640-1280. (The referenced graph shows a phase peak here.)

Interesting observation on the phase plot. I've never paid too much attention to phase. How does phase play into the sound we hear? The peak you mention, which I assume is the small one between 2-3KHz, is much smaller than the broad one centered at 700Hz. What are your thoughts on the larger peak?

BTW - I meant to say that I heard problems in the upper midrange / lower treble of the Focal speakers, which to me is the 1KHz - 4Khz range. I verified that the problem was in the range covered by the tweeter, by placing a sock over the tweeter. With the sock was on, the problem went away.

Around 3k is about where many speakers cross over from mid to tweeter. This can be a problem range since it is also in the range that most people are sensitive to. It is also a range that can easily be changed by the room. It is also a range where personal preference plays a big role. What some people perceive as harshness others might perceive as the proper bite that some instruments have in that range. It is also a range where midrange cone breakup can be a problem. Likely your cloud helped in this regard, it got one of the reflection points (the one most people usually neglect). Had you already traded out speakers before adding the cloud?

Let's see an array of measurements comparable to these

http://www.soundstagenetwork.com/measurements/speakers/paradigm_signature_s1_v2/

before we decide that 'measurements don't explain' hifisponge's perception.

One of the points Toole hammers home is that WHAT we measure matters as much or more than the measurements themselves.

Around 3k is about where many speakers cross over from mid to tweeter. This can be a problem range since it is also in the range that most people are sensitive to. It is also a range that can easily be changed by the room. It is also a range where personal preference plays a big role. What some people perceive as harshness others might perceive as the proper bite that some instruments have in that range. It is also a range where midrange cone breakup can be a problem. Likely your cloud helped in this regard, it got one of the reflection points (the one most people usually neglect). Had you already traded out speakers before adding the cloud?

Yes, I traded out the Focal speakers before the acoustic treatment was installed on the advice of the acoustician. He felt that room treatment would not solve the issue he and I heard with that speaker.

I'm open to the possibility that I'm sensitive to the 3KHz range, but I did some informal testing on this by listening to music through headphones that sound good to me. I would play a piece of music that I found particularly offensive (as heard through the Focals) through the headphones and intentionally goose the EQ in the range from 1K to 8K by as much as 3 to 6dB. Somewhat to my surprise, none of the EQ boosts I made replicated harshness I heard in the Focal speakers.

Let's see an array of measurements comparable to these

http://www.soundstagenetwork.com/measurements/speakers/paradigm_signature_s1_v2/

before we decide that 'measurements don't explain' hifisponge's perception.

One of the points Toole hammers home is that WHAT we measure matters as much or more than the measurements themselves.

The only thing that soundstage measures that Stereophile does not is THD. Well, they also test FR linearity at high output, but that is irrelevant because I heard problems with Focal speaker at moderate volume.

While soundstage has not measured the Focals, I was able to find distortion measurements on the Focal 1037 speaker taken by Audio.de magazine.

http://i80.photobucket.com/albums/j193/ptwalker/focal1037distortion.jpg

You can see that distortion remains low as long as the speaker isn't pushed to 100dB, but the majority of the distortion is clearly below the range that I heard problems. And since I heard the "harshness" in the upper mids / lower treble at volume levels closer to 70-80dB, there is no data that I can find on this speaker that points to the problems I heard.

To the credit of the Stereophile measurements, there was one thing that they measured that I could hear, and was a bit surprised to hear, and that was a peak in the port output in the lower midrange. This showed up when listening to a solo female vocal track as a wooly quality to the lower notes of this singer. When I placed my hand on the side of the cabinet while these low notes were played, the cabinet would light up from the resonance. I was surprised that I could hear and feel this resonance because it seemed to be low enough in magnitude on the graph that I didn't think it would be an issue.

I'm referring to the peak at roughly 400Hz in the red trace.
http://www.stereophile.com/images/archivesart/707Focfig3.jpg

This should be of interest to anyone following this thread.

http://seanolive.blogspot.com/2009/01/what-loudspeaker-specifications-are.html

Here's an excerpt:
...the final hours of the symposium in a panel discussion called: “What loudspeaker specifications are relevant to perception?”

The panelists included myself, Steve Temme (Listen Inc.), Dr. Earl Geddes (GedLee), Laurie Fincham (THX), Mike Klasco (Menlo Scientific), and Dr. Floyd Toole (former VP Acoustic Engineering at Harman), who served as the panel moderator. After about 30 minutes, a consensus was quickly reached on the following points:

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).

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.

4.Current industry loudspeaker specifications are woefully inadequate in characterizing the sound quality of the loudspeaker. The commonly quoted “20 Hz - 20 kHz , +- 3 dB” single-curve specification is a good example. Floyd Toole made the observation that there is more useful performance information on the side of a tire (see tire below) compared to what’s currently found on most loudspeaker spec sheets (see Floyd's new book "Sound Reproduction").

I find point #3 particularly interesting.

I would also like to know how they measure the qualities in the "sound quality attribute" graph in the lower right corner. Which measurements show us the predicted "spaciousness", "lack of coloration", and "dynamics" of a loudspeaker?

It's probably a matter of you and those others putting way too much stock into a thread on the internet. Social failure ftl.

Subjectivists probably enjoy listening to the music, not the systems, which I wonder about with the objective crowd, always turning a hobby into a lab experiment. More to your question's point though, the issue is that more than likely, consumers will not magically follow someone else's experiences as a guide when purchasing themselves. So really, whatever findings are involved are a study in wasting oxygen.




WTF? This is a a science discussion in a science forum. What are you doing posting here? Why even bother with this site, just go listen and enjoy...that is cool.

I can tell you one thing, Im extremely aggresive in helping move people towards a more educated purchase. Friends, family and anyone on the speaker forum. Anytime someone posts "Just Go listen" Im going to reply "That is silly for X reasons.....".

Btw, I think you also do not realize that posting on here has little to do with the enjoyment of speakers for movies or music. I care about Movies, I have an Ipod for music! Most forums on this site have zero to do with sharing enjoyment stories (Like I care if someone enjoys some music :eek: Find a forum where people care!!).

I do not care if people get hard-ons while listening to their weird choices of music. This site is 99% science and product information to me and I treat it as such. I have purchased thousand of dollars worth of equipment off the "for sale" forums (I have sold products too), I have built several HTPCs, bought BD players, Researched media players, tested many AVRs/amps, built my own full house automation system, Built my own custom HT room with DIY screen and all DIY speakers. I have used this site to its maximum potential...what are you doing here is what I wonder? :eek:

Now can we get back on topic without subjectivists trying to derail the idea that science has a HUGE role in speakers.

This should be of interest to anyone following this thread.

http://seanolive.blogspot.com/2009/01/what-loudspeaker-specifications-are.html

Here's an excerpt:


I find point #3 particularly interesting.

I would also like to know how they measure the qualities in the "sound quality attribute" graph in the lower right corner. Which measurements show us the predicted "spaciousness", "lack of coloration", and "dynamics" of a loudspeaker?


Thanks for the link, Those are the measurements I would love to know how to measure.....If we can measure it all it makes it all so easy so that everyone can get to enjoying their purchase quicker then pulling their hair out weeks on end, torturing themselves over any speaker purchase.

This should be of interest to anyone following this thread.

http://seanolive.blogspot.com/2009/01/what-loudspeaker-specifications-are.html

Here's an excerpt:


I find point #3 particularly interesting.

I would also like to know how they measure the qualities in the "sound quality attribute" graph in the lower right corner. Which measurements show us the predicted "spaciousness", "lack of coloration", and "dynamics" of a loudspeaker?
Point #3 is something that Earl was looking at with respect to finding what metric, or appropriately weighted combination of, correlated with preference. There may still be a link on his website regarding some work that he did.

I'll have to get around to reading Sean's blog, Tim, but do you recall what Sean attributed the linear distortions to in a speaker?

Yes, I traded out the Focal speakers before the acoustic treatment was installed on the advice of the acoustician. He felt that room treatment would not solve the issue he and I heard with that speaker.


Well, that makes what you heard even more of a 'black box'. It would have been interesting to see before/after treatment measurements of the room with the Focals.

]
I'm open to the possibility that I'm sensitive to the 3KHz range, but I did some informal testing on this by listening to music through headphones that sound good to me. I would play a piece of music that I found particularly offensive (as heard through the Focals) through the headphones and intentionally goose the EQ in the range from 1K to 8K by as much as 3 to 6dB. Somewhat to my surprise, none of the EQ boosts I made replicated harshness I heard in the Focal speakers.

Suggesting again the possiblity of a room/speaker interaction effect.

This should be of interest to anyone following this thread.

http://seanolive.blogspot.com/2009/01/what-loudspeaker-specifications-are.html

Here's an excerpt:


I find point #3 particularly interesting.

Perhaps because you're casting about for answers that fit your theory (i.e., that there was something 'wrong' with the Focals that wasn't measurable)?

I woudl think #1 and #2 -- the agreement that perception of loudspeaker quality is *dominated* by known, measureable factors , and that listeners tend to prefer accurate loudspeakers -- are at least as interesting...and really rather important findings.



I would also like to know how they measure the qualities in the "sound quality attribute" graph in the lower right corner. Which measurements show us the predicted "spaciousness", "lack of coloration", and "dynamics" of a loudspeaker?


You should get yourself a copy of Floyd Toole's book, and look up Olive's JAES articles.

WTF? This is a a science discussion in a science forum. What are you doing posting here? Why even bother with this site, just go listen and enjoy...that is cool.

I can tell you one thing, Im extremely aggresive in helping move people towards a more educated purchase. Friends, family and anyone on the speaker forum.

This isn't a "science forum" because occasionally science is mentioned. The whole point of any of these discussions is that for most people (and I say this as someone who's boots are on the ground) these studies only carry so much weight. People want accuracy, but long gone are the days when people are taking notes about tech specs.

They listen.

Anytime someone posts "Just Go listen" Im going to reply "That is silly for X reasons.....".

That's appreciated. But at the point of purchase, it's forgotten.

Now can we get back on topic without subjectivists trying to derail the idea that science has a HUGE role in speakers.

That's putting words in my mouth. Science's role is in the proper designing of the speaker. Personal subjective values tend to be applied when people PURCHASE the speaker. Your proposal apparently is to remove things like demo rooms and facilities to experience the product, instead only worrying about the specifications.

This isn't a "science forum" because occasionally science is mentioned. The whole point of any of these discussions is that for most people (and I say this as someone who's boots are on the ground) these studies only carry so much weight. People want accuracy, but long gone are the days when people are taking notes about tech specs.

They listen.


They only carry so much weight because many are stubborn and refuse to let science get in the way of a sale. I read it all the time in several forums. The "I let my ears decide everything" Crowd try to always argue against science. They are the 'Belief' society and its also just easier to be that way so people are easily convinced.



That's appreciated. But at the point of purchase, it's forgotten.


Thats fine, as long as I have posted what I believe a certain percent of people will learn. You can not teach everyone but we still have to keep trying. Again, I have no problem with people buying what they like even if its just because it looks great...I have no problem at all!


That's putting words in my mouth. Science's role is in the proper designing of the speaker. Personal subjective values tend to be applied when people PURCHASE the speaker. Your proposal apparently is to remove things like demo rooms and facilities to experience the product, instead only worrying about the specifications.

What I read from you in this thread and other threads is that you are a strong supporter of simple subjective listening, you do not like the same opinions posted over and over and over when it comes to pointing out the "snake oil" or "Placebo induced" decision problems so many people make. I have no idea if you want to educate anyone to help them figure things out better but I suspect you want less science debate more "How did it sound to you" posts.

btw, demo rooms/facilities are a complete waste of time IMO. The proper place to make a logical decision on any set of speakers is in your own room, with your own equipment, etc. Its completely foolish to think that any demo will remotely match what you have in house. local HT shops are dying a slow death and Im okay with that because it gets us to a place where I think Measurements/specs will actually have a lot more meaning and that is great thing from my perspective.

Clearly, the only proper way to evaluate loudspeakers is to listen to them a system at a time using music that you like, and then to jot down all the adjectives and phrases that come to mind over several weeks, even when you're away from the loudspeakers.

Repeat the process with other loudspeakers you wish to evaluate.

At the end, compare the adjectives and phrases associated with the different loudspeaker systems. Numerically rank the more flowery or emphatic ones higher, then add them up to create a "Florid and Torrid" scale. The loudspeaker system that scores the highest is the best.

This method works best if you announce your results next Wednesday or its anniversary.

Perhaps because you're casting about for answers that fit your theory (i.e., that there was something 'wrong' with the Focals that wasn't measurable)?

I woudl think #1 and #2 -- the agreement that perception of loudspeaker quality is *dominated* by known, measureable factors , and that listeners tend to prefer accurate loudspeakers -- are at least as interesting...and really rather important findings.

You should get yourself a copy of Floyd Toole's book, and look up Olive's JAES articles.

I believe that what I heard with the Focals can be measured, as it was so obvious that some sort of measurement would have to be able to read it. The sound was not subtle. I’m just stating that based on my knowledge of reading the common measurements (FR, THD, and Spectral Decay), I wasn’t able to connect what I heard with any of these. I am also open to the possibility that the problem with those speakers could be related to FR, I just find it unlikely based on my own measurements. But then, I am self taught, so it could be that I’m looking in the wrong spot. Which comes back to the issue of which is easier, learning how to interpret speaker measurements, or just listening? If listeners “tend to prefer accurate speakers”, then when we hear a more accurate speaker we should naturally choose it over a less accurate one, so why bother with the measurements?

I only pointed out bullet #3 in the Olive article because it acknowledges that there are still some unknowns, which is the point of this thread.

I have read Toole’s early whitepapers on how test subjects preferred flat FR, but I’m not up to speed on his current works, or Olive’s. I’ll have to read up on those before I get too much deeper into the subject of this thread.

This isn't a "science forum" because occasionally science is mentioned. The whole point of any of these discussions is that for most people (and I say this as someone who's boots are on the ground) these studies only carry so much weight. People want accuracy, but long gone are the days when people are taking notes about tech specs.


If that's true (and on AVSF, I doubt that it is) , maybe it's time they came back. Bench tests these days can be pretty frickin' awesome for data mining. We're even on our way to predicting how good a loudspeaker sounds, from the 'tech specs'.

They listen.


And they often make silly claims based on that.


That's putting words in my mouth. Science's role is in the proper designing of the speaker.


And in studying the relationship of design to what people think 'sounds good' -- no?

If you don't think so, too bad: the genie's already out of the bottle.

Clearly, the only proper way to evaluate loudspeakers is to listen to them a system at a time using music that you like, and then to jot down all the adjectives and phrases that come to mind over several weeks, even when you're away from the loudspeakers.

Repeat the process with other loudspeakers you wish to evaluate.

At the end, compare the adjectives and phrases associated with the different loudspeaker systems. Numerically rank the more flowery or emphatic ones higher, then add them up to create a "Florid and Torrid" scale. The loudspeaker system that scores the highest is the best.

This method works best if you announce your results next Wednesday or its anniversary.

Don't forget to add a box for 'wife's comments'. Because we know something isn't TOTALLY true until it's got 'even my wife heard it' evidence.

I believe that what I heard with the Focals can be measured, as it was so obvious that some sort of measurement would have to be able to read it. The sound was not subtle. I’m just stating that based on my knowledge of reading the common measurements (FR, THD, and Spectral Decay), I wasn’t able to connect what I heard with any of these. I am also open to the possibility that the problem with those speakers could be related to FR, I just find it unlikely based on my own measurements. But then, I am self taught, so it could be that I’m looking in the wrong spot. Which comes back to the issue of which is easier, learning how to interpret speaker measurements, or just listening? If listeners “tend to prefer accurate speakers”, then when we hear a more accurate speaker we should naturally choose it over a less accurate one, so why bother with the measurements?


Well, Atkinson's measurements did show some things (and didn't show others -- a set of off-axis FR curves in Fremer's room could have been informative), like beaming above 12kHz, a resonance at 344 Hz (possibly causing 'midrange congestion at high levels'), and a slight hump in the 4-11k region.

Those are gorgeous loudspeakers, btw. You must have been sad to part with them.


I only pointed out bullet #3 in the Olive article because it acknowledges that there are still some unknowns, which is the point of this thread.

I don't think anyone disputed that. I just suspect points #1 and #2 were given precedence over #3 for a reason, in Sean's summary.


I have read Toole’s early whitepapers on how test subjects preferred flat FR,


not just flat on-axis, but (surprisingly) flat OFF-AXIS too. I think it's important to always mention that.

but I’m not up to speed on his current works, or Olive’s. I’ll have to read up on those before I get too much deeper into the subject of this thread.

The "I let my ears decide everything" Crowd try to always argue against science.
But it's not exactly science they are arguing against. It is not a scientific fact or a scientific theory that one should buy a speaker with a flat FR, e.g. Science doesn't furnish "should"s. It may be a fact that a certain speaker has a flat response, but so what? How do such facts tell people what speakers are worth or which ones they should buy?

I later hired an acoustician to evaluate my room and to check my sanity with the sound I was hearing, and he concluded that the problem in the upper mids was a character of the speaker, not the room.

A second opinion from a pro is always useful. I see a 4 dB rise centered at 5 KHz in your graph. Another possibility is that the tweeter is ringing. Ringing is not reduced with EQ, only the amplitude. I also agree with dknightd that absorption at the reflection points will probably help a lot, even if you're very close to the speakers. Unless the side walls and ceiling are all at least ten feet away.

--Ethan

Those are gorgeous loudspeakers, btw. You must have been sad to part with them.


It was very hard indeed. Still one of the best looking speakers out there, IMO. I special ordered them in a Macassar Ebony finish and waited 3 months for them to arrive.

I did listen to them first at a local showroom, and looking back at that I should have heeded to the warning signs, but I thought that they were just "revealing". Living with them for three months proved different.

I tried everything within my power to make them work, but in the end I just couldn't correct or get past the issue I heard in the upper mids / lower treble.

But it's not exactly science they are arguing against. It is not a scientific fact or a scientific theory that one should buy a speaker with a flat FR, e.g. Science doesn't furnish "should"s. It may be a fact that a certain speaker has a flat response, but so what? How do such facts tell people what speakers are worth or which ones they should buy?

That is pretty well the point of this thread is to find out what measurements are still needed to make our decisions easier.

Mark and others have pointed out that a flat FR can still sound different from driver to driver...I already know that because I have lots of drivers in house these days and I can use my DCX to create a flat FR for each of them and they will still sound different. My question I keep wondering then what measurement will show me that difference?

This in the end does not stop people from going out and just enjoying shopping for speakers. Im not try to force anything on anyone. I just believe there is room for more data wrt our purchase decisions.

How do you measure: Spaciousness?


Interaural correlation in ERB's.


Size of Soundstage?


Principle axis analysis


Warmth?


Spectrum


Bloated Sound?


Direct to reverberant ratio, spectrum.


Muddiness?


Bass response


Detail?


Amount of preemphasis.


Openess?


Speaker radiation pattern, interaural correlation, interchannel time delay.


Boxiness?


Frequency response.


Accuracy is just a small part of the music equation.


Define "accuracy" for me, please.

Very true so why does 95% of the population pick from 100s of different brands?

People prefer different direct to reverberant patterns in the listening room.

That really is really the next issue beyond frequency response, I think.

Interaural correlation in ERB's.

Principle axis analysis

Spectrum

Direct to reverberant ratio, spectrum.

Bass response

Amount of preemphasis.

Speaker radiation pattern, interaural correlation, interchannel time delay.

Frequency response.



Define "accuracy" for me, please.

Good to see a professional in the audio field chime in. Your response is thought provoking, if a bit terse. Care to elaborate? What is ERB? And can a speaker "pre-emphasize" the signal it is being fed? Seems like that would require some sort of signal processing. And what is preemphasis? What do you look for in the "principle axis analysis" to predict the size of the soundstage? Why is the sky blue? Are we there yet? :D

Your response is thought provoking, if a bit terse. Care to elaborate?

That quickly leads to a graduate-level seminar if not a whole course or series of them.


What is ERB?


equivalent rectangular bandwidth


And can a speaker "pre-emphasize" the signal it is being fed?


They all do that. ;-)


Seems like that would require some sort of signal processing.


They all do that. ;-)


And what is preemphasis?


Usually, some kind of non-flat frequency response.


What do you look for in the "principle axis analysis" to predict the size of the soundstage?


"principle axis analysis" is a kind of statistical analysis that can be applied to sound field data to characterize its spatial nature.

People prefer different direct to reverberant patterns in the listening room.

That really is really the next issue beyond frequency response, I think.

That one is way over my head :eek:

Not sure about Tom Holman con this Auduesy or how ever its spelt I don’t care not interested it sounds like a load of hot air.

Have you ever listened to a properly set up Audyssey based system? Do you have any clue as to what Room Correction is? Based upon your comment here I would have to suspect the answer to both of these questions is "No".

I have an Audyssey enabled Integra and the Pro kit and I can tell you first hand that it is anything but a con. I know from experience and an understanding of what it does that it is a very real benefit and far exceeds pretty much anything else you can do. In terms of value there is nothing else I can think of that provides as much value for the buck.

The irony here is on many levels. First, your comments in the rest of your posting amount to a crude attempt at doing what Audyssey does in a far more sophisticated manner.

Second, when using a system such as Audyssey, you reduce the impact of Speaker and Room induced distortions. While it will not produce absolutely perfect sound it does, in most if not all cases, make a massive improvement in the quality of the experience when done properly. If I turn the Audyssey system off the difference is like night and day and is audible to anyone that is not stone deaf. If you are going to quibble about the sonic differences in any component other than speakers then you have to try Audyssey as the impact on the sound exceeds pretty much anything else you can do.

I would hazard the claim that decent speakers in a room with Audyssey correction will sound better than fine speakers in the same room without Audyssey in most cases. The room distortion in most situations is so signficant that few other improvements can come close, and if the speaker does have some issues those can often be addressed as well.

I know this from experience and from my understanding of the technology. If you are going to make a statement like you made here I suggest you actually study the system and experience it before inserting your foot in your mouth.

I will close by pointing out that our company uses similar techniques to equalize the response of our audio systems which go into high end business jets. While not exactly the same methods, many of the same principles are used. Audyssey is a very real and effective technology.

Have you ever listened to a properly set up Audyssey based system? Do you have any clue as to what Room Correction is? Based upon your comment here I would have to suspect the answer to both of these questions is "No".You should understand who you're replying to:

http://archive2.avsforum.com/avs-vb/showthread.php?p=8979768&&#post8979768

jj_0001 (James),

Your list is a reasonable first approximation, but of course the devil is in the details. Psychoacoustic qualities never seem to quite line up one-to-one with measured quantities. See for example, Lavandier et al., "Identification of some perceptual dimensions underlying loudspeaker dissimilarities," J. Acoust. Soc. Am. 123 (6), June 2008. This paper also contains a very nice literature review.

Add to this a healthy dose of linguistic ambiguity, which of course depends on specific language. I'm thinking of such studies as Ozawa et al., "Psychological factors involved in auditory presence," Acoust. Sci. & Tech. 24, 1 (2003). This research was specific to one particular Japanese sound word!

I believe that everything in sound perception is measurable with current instrumentation. But constructing accurate psychoacoustic correlates and models from this data is another matter, and one which will likely keep researchers busy for decades to come.

Regards,
Terry

You should understand who you're replying to:

http://archive2.avsforum.com/avs-vb/showthread.php?p=8979768&&#post8979768

Ah. I see. Yes, I forgot that there are folks like that out there. Thank you very much for pointing this out, I will not expect a rational response after seeing that... ;)

I do appreciate your bringing this to my attention.

That quickly leads to a graduate-level seminar if not a whole course or series of them.


Sounds like fun to me :) I wonder who would teach it. Maybe team taught.
What should the prerequisites be?

Sounds like fun to me :) I wonder who would teach it. Maybe team taught.
What should the prerequisites be?
MIT is not strong in psychoacoustics, but this might be a place to start (and the price is right!):
6.551J / HST.714J Acoustics of Speech and Hearing (http://ocw.mit.edu/OcwWeb/Electrical-Engineering-and-Computer-Science/6-551JFall-2004/CourseHome/index.htm)

- Terry

MIT is not strong in psychoacoustics, but this might be a place to start (and the price is right!):
6.551J / HST.714J Acoustics of Speech and Hearing (http://ocw.mit.edu/OcwWeb/Electrical-Engineering-and-Computer-Science/6-551JFall-2004/CourseHome/index.htm)

- Terry

Nice. Looks like there might some interesting reading material there :) It is a few years old, but still likely useful. Thanks for the pointer

jj_0001 (James),

Your list is a reasonable first approximation, but of course the devil is in the details. Psychoacoustic qualities never seem to quite line up one-to-one with measured quantities. See for example, Lavandier et al., "Identification of some perceptual dimensions underlying loudspeaker dissimilarities," J. Acoust. Soc. Am. 123 (6), June 2008. This paper also contains a very nice literature review.

Add to this a healthy dose of linguistic ambiguity, which of course depends on specific language. I'm thinking of such studies as Ozawa et al., "Psychological factors involved in auditory presence," Acoust. Sci. & Tech. 24, 1 (2003). This research was specific to one particular Japanese sound word!

I believe that everything in sound perception is measurable with current instrumentation. But constructing accurate psychoacoustic correlates and models from this data is another matter, and one which will likely keep researchers busy for decades to come.

Regards,
Terry

I don't think it's quite that grim, but of course you've a very good set of points.

The problem comes from my last question, that being "what is accuracy". Given that when we record in an acoustic space, for instance, we capture about a bazillionth of the actual information in the sound field (most of which the ear could never resolve, btw), first we need to decide what we call "accuracy'.

Then we need to actually define terms like spatiality, etc, and then do a PEA to see how our terms fit to our measures using our definitions, and do another fit of the measures, and the definitions, to the common lexicon.

Yeah, I agree, it's not perfect, I certainly wouldn't argue that, but I think that the poster I replied to was arguing that science was helpless, and it certainly isn't.

That quickly leads to a graduate-level seminar if not a whole course or series of them.


I'm afraid Arny is kinda right here. What one CAN measure, vs. what the industry usually measures, is a very long story. And it would be a series of courses, starting with "how the ear works".

A painfully condensed version can be found at www.aes.org/sections/pnw/ppt.htm but I admit that having the lecture to go along with it would help an awful lot.

That one is way over my head :eek:

Sorry, let me take it a bit slower.

Any loudspeaker has a radiation pattern that varies with angle, frequency, distance, etc.

This radiation pattern interacts with the acoustics of the room that the speaker is operating in.

The result is a particular direct to reverberant characteristic (both frequency and time) due to the combination.

Different people prefer different characteristics, and as such, different loudspeakers, different rooms, etc.

Sounds like fun to me :) I wonder who would teach it. Maybe team taught.
What should the prerequisites be?

Well, a good understanding of frequency analysis, basic signal processing, and basic acoustics for starters.

You'll need the frequency analysis to understand time/frequency tradeoffs, the signal processing to be able to formalize how the ear works, what speakers do, how to measure signals, etc, and basic acoustics to build on how the ear interacts with the soundfield, how a room's impulse response looks, and how it came to be that way, etc.

I'd love to teach such a course, but it would have to be grad-level, or the prereq's would take a year or two.

Can't you encompass the room in binaural recordings?

Ultimately sound reproduction is an art and with that comes subjectivity. Evidenced by the extravagent prices people pay for their equipment. That is the differentiating factor between art and utility. People don't pay for a priceless nail but a Da Vinci can be priceless.

If people paid strictly for utility then there wouldn't be such a giant range of prices. The range is so large it is in magnitudes($100 to $100000). Conversely take the price of bread. We subjectively like the "taste" of bread but the range of prices for a loaf are between $2 and $8. Bread is not a form of art.

A significant factor of art is that we pay more for what we don't see. Compare market values of impressionist paintings to a literal Norman Rockwell to a very literal leibowitz photograph. The value we place on them is a combination of the actual thing you get, what you see, but mostly what you think the creating artist put into the work. that there is something deeper into the work that you just can't see.

Musical art is no different. My really favourite stuff can be revisited years later and I seem to hear different parts I didn't before. It was always there but I didn't notice it or wasn't sophisticated enough to appreciate it.

Since audio reproduction is subjective the science is only valuable if you have experience with your own tastes and can couple it with theoretcal understanding.

But it's not exactly science they are arguing against. It is not a scientific fact or a scientific theory that one should buy a speaker with a flat FR, e.g. Science doesn't furnish "should"s. It may be a fact that a certain speaker has a flat response, but so what? How do such facts tell people what speakers are worth or which ones they should buy?

Well, flat direct response? Flat power response? Flat both? What flat? At what distance, what angle?

There are many variables to loudspeakers.

Again, whenever you have THD you also have IMD. Like peas in a pod.

Well, IMD and THD are two aspects of nonlinearity, and you certainly must have one when you have the other.

The only real difference is that with THD you're looking at the results of one sinusoid, and with IMD at least two. But the mechanisms that give rise to one necessarily give rise to the other.

Of course, since everything has differently linearity at different frequencies, you may not see this as clearly in measurements as you might expect. I suspect a treatise on powers of (e^jw - e^-jw) (sinusoid) is not called for, let alone powers of a combination of several different 'w's.

Ultimately sound reproduction is an art and with that comes subjectivity. ...
Since audio reproduction is subjective ...
I disagree with this idea. Audio equipment preference is subjective. Visual design, branding, price, fine Corinthian leather, etc. all figure into this. But equipment with perfect accuracy of reproduction should sound the same. Audible differences, as determined by double blind testing, must then be reduced to distortion.:eek: This begs the "what is accuracy?" question raised by jj_0001.

Regards,
Terry

I disagree with this idea. Audio equipment preference is subjective. Visual design, branding, price, fine Corinthian leather, etc. all figure into this. But equipment with perfect accuracy of reproduction should sound the same. Audible differences, as determined by double blind testing, must then be reduced to distortion.:eek: This begs the "what is accuracy?" question raised by jj_0001.

Regards,
Terry

+1 I was wondering how to word the subjectivity...your "equipment preference is subjective" is perfect.

I disagree with this idea. Audio equipment preference is subjective. Visual design, branding, price, fine Corinthian leather, etc. all figure into this. But equipment with perfect accuracy of reproduction should sound the same. Audible differences, as determined by double blind testing, must then be reduced to distortion.:eek: This begs the "what is accuracy?" question raised by jj_0001.

Regards,
Terry

Kinda off topic a bit but do you always want "accurate" sound? The pursuit of "good sound" and "accurate sound" are not the same. I would love an audio processor that would magically clean up Nina Simone records to something of FM radio broadcast quality. If it "distorted" the source, I wouldn't care so long as it sounded better. There is something lost in the voice of a Simone recording vs an Amy Winehouse recording. Can we measure the differences and add more to one and less to another? Not with todays technology but someday.
I don't believe our preferences will ever lead to a limit. Just a tight statistical deviation. I will always prefer a little less bass and I imagine I will always prefer less dynamic range (than those in this community). A basshead will always pound out the subs driving their riced up car. That is their preference of "good sound". It is no more wrong than what a measurement will dictate.

Science is an invaluable tool. I am not discounting it's worth. But the last mile..the part from the mp3 to my inner ear, that's all me and my preference. Science doesn't measure emotion or passion. They are critical to the pursuit of good sound.

But it's not exactly science they are arguing against. It is not a scientific fact or a scientific theory that one should buy a speaker with a flat FR, e.g. Science doesn't furnish "should"s. It may be a fact that a certain speaker has a flat response, but so what? How do such facts tell people what speakers are worth or which ones they should buy?


Well, flat direct response? Flat power response? Flat both? What flat? At what distance, what angle?

There are many variables to loudspeakers.
My question was about "such facts", so it would be appropriate to consider any of the measures you mention, refined and elaborated in any way you think suitable, if you'd like to tackle the question I asked. I'm skeptical that there is any principled way to get from the science to the recommendations.

Kinda off topic a bit but do you always want "accurate" sound? The pursuit of "good sound" and "accurate sound" are not the same. I would love an audio processor that would magically clean up Nina Simone records to something of FM radio broadcast quality. If it "distorted" the source, I wouldn't care so long as it sounded better. There is something lost in the voice of a Simone recording vs an Amy Winehouse recording. Can we measure the differences and add more to one and less to another? Not with todays technology but someday.
I don't believe our preferences will ever lead to a limit. Just a tight statistical deviation. I will always prefer a little less bass and I imagine I will always prefer less dynamic range (than those in this community). A basshead will always pound out the subs driving their riced up car. That is their preference of "good sound". It is no more wrong than what a measurement will dictate.

Science is an invaluable tool. I am not discounting it's worth. But the last mile..the part from the mp3 to my inner ear, that's all me and my preference. Science doesn't measure emotion or passion. They are critical to the pursuit of good sound.


You can separate the two.....measurement/science will give you a very good idea if a speaker is good or not. measurements should also give you an idea about how the speakers sound. This again is a hypothetical point because we all know many measurements are not accurate or they just are not done.

As for people wanting an accurate or inaccurate sound that should have little to do with speaker purchases...HONESTLY people shoulld buy speakers based on how well they are built, how well they measure, how much they cost (budget concerns) and how few flaws their sound has. Once the right purchase is made people should take that scientifically accurate/well built speaker into their room and EQ it to the sound they want!

Chasing a specific sounding speaker (inaccurate/distorted response) only to have it change in our rooms because of Audyssey, in room problems, System setup issues, etc makes me wonder daily how much people honestly do not care about the science and constantly convince themselves they are hearing something no matter what room the speakers are in.

More to the point, you do not find a speaker with more bass, you EQ your system to give you more bass...do you understand that? If you did choose well build/accurate speakers this is much easier to do....5 years down the road your preferences change, you do not buy new speakers you once again EQ to your current needs.

penngray,
I'm going to turn this around. You seem to like measurements. You've measured the speakers you've built. You like they way they sound.

What part of the measurements do you think is the most important?

Have you ever tried comparing your DIY speakers with something that can be purchased ready made? Have you done this blind? Do measurements tell you
which is better?

In a sighted test a DIY speaker will always be better - because you built it yourself.

If you compared two speakers based only on measurements, what would you look for in those measurements? If you had two speakers that seemed to measure the same do you think you could tell them apart - and more important, tell which was which without looking?

But equipment with perfect accuracy of reproduction should sound the same.


Except that one of the important aspects of speakers in listening rooms is to attempt to replace or simulate, somehow, what was lost in the original recording by the simple fact that most of the information is necessarily lost in conversion to 2 channels of audio. Or 5, or 7. And different people may prefer different "replacements". So accuracy is reduced, I think, to something involving electrical inputs as they compare to speaker output. There are still many variables in radiation pattern, etc, to be dealt with.


Audible differences, as determined by double blind testing, must then be reduced to distortion.:eek: This begs the "what is accuracy?" question raised by jj_0001.

Regards,
Terry

Not necessarily distortion. Distortion is generally used to describe nonlinear signal modifications. Things like frequency shaping, differences in direct vs. reverberant timbre in the playback room, etc, are not generally (leaving out buzzing, rattling lights, etc) nonlinear, but they surel cause differences.

Those who are arguing the issue ought to read Floyd's book, I think. It gets into many of the issues, and while I don't entirely agree with it, it's a good start for speakers that don't deliberately have dipole/bipole radiation patterns.

I'm skeptical that there is any principled way to get from the science to the recommendations.

"recommendations" require some knowlege of the listener's preferences in distortions, frequency shaping, and direct to reverberant ratio for listening at the very least.

You are leaving out the need to evalute the listener for the listener's preferences, and you can't do that. Preferences vary widely, perhaps even wildly. Some 30% or so prefer very diffuse sound, and go for a particular kind of speaker. Some others prefer almost all direct. Most us not in those two groups want different direct/diffuse combinations for different music.

So, first, know your preferences. It is likely you can relate them to actual measurements to start with, if given the right information on setups that you listen to. THAT information is just about always 100.00% absent. :(

Well, flat direct response? Flat power response? Flat both? What flat? At what distance, what angle?

There are many variables to loudspeakers.

Thank you.

The problem is not measuring speakers (see NWAA Labs & ETC Inc). The problem lies in the complexity of assembling the data which can be gathered into a single representation of what we will hear.

The notion of a fully comprehensive set of measurements which wouldn't miss issues makes my head spin if you want to get really detailed. In most cases the most telling measurements are reasonable in number, but come from investigative work based on observations of other measurements. The talk of distortion and IMD is a good example, as it is easy to move weaknesses away from the test ranges until you get into full transfer function measurements, which then have the problem of being very hard to interpret visually and are generally 3-4D problems.

I do see some strong potential in convolution with proper binaural auralization. Tools for this and further convolution with a carefully modeled acoustic environment already exist. The biggest limitation here is that you cannot capture any dynamic changes in behavior of a loudspeaker, although a complex enough model with detailed measurements and auralization options could at least let you hear the character of a loudspeaker, potentially approximated in your own room.

Except that one of the important aspects of speakers in listening rooms is to attempt to replace or simulate, somehow, what was lost in the original recording by the simple fact that most of the information is necessarily lost in conversion to 2 channels of audio. Or 5, or 7. And different people may prefer different "replacements". So accuracy is reduced, I think, to something involving electrical inputs as they compare to speaker output.

Yes, that was my meaning. The same electronic signal, however it was recorded, as reproduced by different speakers in the same acoustical environment.

There are still many variables in radiation pattern, etc, to be dealt with.

Yes, as there is no standardized reproduction environment for 2-channel audio, as there is for motion picture multi-channel sound. But I think it can be generally agreed* that flat frequency response, both on and off axis, are minimally necessary for accurate reproduction. That is straightforward to measure and plot, and relatively easy to interpret visually.

Not necessarily distortion. Distortion is generally used to describe nonlinear signal modifications.
Not for the most general meaning of the term. It can include many linear phenomena, such as frequency and phase distortion.

*And confirmed by Sean Olive in his excellent paper "Differences in Performance and Preference of Trained versus Untrained Listeners in Loudspeaker Tests: A Case Study," J. Audio Eng. Soc., Vol. 51, No. 9, 2003 September.

Things like frequency shaping, differences in direct vs. reverberant timbre in the playback room, etc, are not generally (leaving out buzzing, rattling lights, etc) nonlinear, but they surel cause differences.
Sorry, I missed reading this initially. Yes, however such things as direct vs. reverberant timbre can and should be dealt with by the instroduction of standardized listening conditions and the publishing by manufacturers of (already standardized) polar speaker response measurements.

You can separate the two.....measurement/science will give you a very good idea if a speaker is good or not. measurements should also give you an idea about how the speakers sound. This again is a hypothetical point because we all know many measurements are not accurate or they just are not done.

As for people wanting an accurate or inaccurate sound that should have little to do with speaker purchases...HONESTLY people shoulld buy speakers based on how well they are built, how well they measure, how much they cost (budget concerns) and how few flaws their sound has. Once the right purchase is made people should take that scientifically accurate/well built speaker into their room and EQ it to the sound they want!

Chasing a specific sounding speaker (inaccurate/distorted response) only to have it change in our rooms because of Audyssey, in room problems, System setup issues, etc makes me wonder daily how much people honestly do not care about the science and constantly convince themselves they are hearing something no matter what room the speakers are in.

More to the point, you do not find a speaker with more bass, you EQ your system to give you more bass...do you understand that? If you did choose well build/accurate speakers this is much easier to do....5 years down the road your preferences change, you do not buy new speakers you once again EQ to your current needs.


Most people don't care about how a system measures. Frankly I don't care how a system measures. Take any of your favourite albums. Are you listening to the music or listening to the recording? What the measurement brings is the the flexability of the sound system to sound good for a variety of recordings.
Say for example I only listened to solo violin recordings. So I built a speaker inside a Strad. Then every recording I heard sounded like it came from a Strad. Is that not the most ass kicking speaker? How crappy would it measure? The manufacturers mold and react to what the market demands. It would be a travesty if we were forced to purchase "flat" speakers.

To seriously answer your orignal question I need to modify it.
What part of musical sound is not measureable?
Toe-tapping factor
Head banging factor
Emotion
Passion

What part of movie sound is not measureable?
Excitement
"I was there" factor

Certainly if your amp and speaker and room are really bad any part of these elements is lost and on a nice system they are enhanced. When I look at a brochure for a new amp this is what I'ld like to see.

You are leaving out the need to evalute the listener for the listener's preferences, and you can't do that.
You are treating the recommendation problem as though it were the scientific problem of predicting what the listener will choose. I understand why a scientist would do this, but I think it misses the point that most engages us consumers. If I, as a layman, ask you, as an expert, to tell me what the best speaker is, I probably don't mean that I want you to predict what I will buy. Instead, I want you to tell me what's best -- what I should buy.

Say for example I only listened to solo violin recordings. So I built a speaker inside a Strad. Then every recording I heard sounded like it came from a Strad. Is that not the most ass kicking speaker? How crappy would it measure? The manufacturers mold and react to what the market demands. It would be a travesty if we were forced to purchase "flat" speakers.

If by a "strad" you mean a stradivari violin.... then i would suggest that a flat speaker would be better. Its similar to those omni directional speakers that mean to spread sound in many directions to create the reverb, to "make you feel you're there" at the hall of the recording. But all that does, and all your strad box would do is in effect double on the existing effects and harmonics that were captured by the recording. The recordings would then feel exaggerated and unrealistic. That is, unless this part was meant in jest, in which case please excuse me.

Most people don't care about how a system measures. Frankly I don't care how a system measures. Take any of your favourite albums. Are you listening to the music or listening to the recording?
Fidelity to the music is truly not measurable. But there are related matters that are potentially measurable, but are commonly not measured, such as fidelity to the performance or (for HT) the event. In any case, measuring fidelity to the recording is, I agree, of no intrinsic interest.

Most people don't care about how a system measures. Frankly I don't care how a system measures.
Maybe you should. Some good reasons have been presented in this thread.

Say for example I only listened to solo violin recordings. So I built a speaker inside a Strad. Then every recording I heard sounded like it came from a Strad.
No, it wouldn't. It would sound like a Strad sitting inside a Strad. Not the same thing at all.

To seriously answer your orignal question I need to modify it.
What part of musical sound is not measureable?
Toe-tapping factor
Head banging factor
Emotion
Passion
None of these are part of a "musical sound." All are mental reactions to a musical sound. That's a key distinction.

What part of movie sound is not measureable?
Excitement
"I was there" factor
Ditto.

When I look at a brochure for a new amp this is what I'ld like to see.
Then you're looking in the wrong place. Try a fMRI.

Maybe you should. Some good reasons have been presented in this thread.

For the most part the measurements have less value to me as I have little experience to couple them with. I'm not saying they have little value to others.


No, it wouldn't. It would sound like a Strad sitting inside a Strad. Not the same thing at all.

Then to qualify, a speaker designed around a strad to excite its resonances and create the illusion it is being played. Maybe put a bunch of motors on the bridge and excite them with heavy signal processing.


None of these are part of a "musical sound." All are mental reactions to a musical sound. That's a key distinction.

Then how is the "saddness" of the performance transmitted to me without any sound?

Then how is the "saddness" of the performance transmitted to me without any sound?
I want to be glib and say it isn't—that you listen to the performance, and you decide it sounds sad to you. But of course it may well be that the performer is trying to convey "sadness," and we have conventions for doing that which you interpret as the performer intends you to. But there's no guarantee that you'll interpret something the way the performer intended it, nor would you be in any way "wrong" to interpret it differently. So to say that the emotions you feel as you listen to music are "transmitted" to you seems wrong to me.

But let's assume that you and the performer are on the same wavelength, and you get what he's driving at. How does he do it? Well, he controls aspects of the sound—loudness, tempo, timbre, etc. All of those things are definitely measurable. And therefore we can measure the extent to which an audio system preserves those elements so that you might interpret them as intended.

Fidelity to the music is truly not measurable.


You seem to be hung up on answering the wrong question.

However, if you want to go there, fidelity to the music is not even an issue. After they get the basics right peforming music is about a musican making the music his own, which is the opposite of fidelity. Musicans who play with perfect fidelity to the music as written are often called "hacks".


But there are related matters that are potentially measurable, but are commonly not measured, such as fidelity to the performance or (for HT) the event.


That's not a related matter on a forum like this, it is the whole show.


In any case, measuring fidelity to the recording is, I agree, of no intrinsic interest.

As confused at several basic levels as this post it, well yes! :-(

You are treating the recommendation problem as though it were the scientific problem of predicting what the listener will choose.


Predicting what the listener will prefer is of vital interest to many people, not the least of whom is the listener.

As far as applying science to the interests of the listener, why would you have a problem with that, perhaps other than some problem with understanding science?


I understand why a scientist would do this, but I think it misses the point that most engages us consumers. If I, as a layman, ask you, as an expert, to tell me what the best speaker is, I probably don't mean that I want you to predict what I will buy. Instead, I want you to tell me what's best -- what I should buy.

You're clearing making things more complex than they are. In fact both questions are of interest, what you will buy and what you should buy. Ideally they would be the same.

Part of the process of audiophile education is getting the two to converge.

Most people don't care about how a system measures.


I see no evidence that you have some unique qualifactions to speak for most audiophiles. You don't know what most people care about. You can only speak for yourself. You shouldn't blow yourself up like this in public - its embarassing for the rest of us.


Frankly I don't care how a system measures.


You probably only think that way because you haven't thought things out. Presumably you do care what the system sounds like and if you understood measures, you'd understand that there is a strong relationship between how a systems measures and how it sounds.


Take any of your favourite albums. Are you listening to the music or listening to the recording?


People who listen primarly to the music could in general care less what the recording sounds like. It doesn't take a million dollar stereo system to convey musical values. It doesn't even take a $20 stereo to convey musical values. A $19.95 CD player with the $0.05 earphones that come with it will do a wonderful job of conveying musical values. The beat, the pitch, even all of the signficant musical nuances will be there.

So, talking about listening to the music is completely out of place in a discussion of high quality audio products.


What the measurement brings is the the flexability of the sound system to sound good for a variety of recordings.


..or any of them.


Say for example I only listened to solo violin recordings. So I built a speaker inside a Strad.


That's completely weird. Why would you want a strad squared?



Then every recording I heard sounded like it came from a Strad.


Wrong. none of them would sound like they came from a strad. They'd sound like they came from a pretty crappy speaker.

What you are talking about is building a speaker using the case of a strad as its sounding-board instead of a normal diaphragm. In general, musical instrument cases don't make good replacements for the cone of a speaker. Speaker cones are designed to have uniform response to vibrations applied to them, but musical instruments most definately aren't designed to have uniform response to vibrations that are applied to them. Musical instruments are designed to emphasize a certain range of frequencies at the expense of all others. Otherwise, a flute would look like a bass tuba.


Is that not the most ass kicking speaker?


Not at all. It's one of the sillier things I've ever head of - probably the thinking of someone who understands very little about either musical instrument or loudpspeakers.


How crappy would it measure?


Depends on whether you were comparing it to the standards for a speaker or a violin case.

But, its never going to make every violin recording sound like a strad. It's going to be horrid!


The manufacturers mold and react to what the market demands. It would be a travesty if we were forced to purchase "flat" speakers.


It would not be nearly the travesty of a loudspeaker that used a violin case for a diaphragm. The only thing that such a speaker would make sound like a violin, would be a special recording made of the vibrations of the bridge of a violin.


To seriously answer your orignal question I need to modify it.
What part of musical sound is not measureable?


Again confusion seems to reign in your post. The following are attributes of music that can easily transcend the sound quality of the equipment reproducing it, once that equipment meets certain very minimal standards:



Toe-tapping factor
Head banging factor
Emotion
Passion
What part of movie sound is not measureable?
Excitement
"I was there" factor

Certainly if your amp and speaker and room are really bad any part of these elements is lost


That would be really, really bad. Fact is that all of the above are well-reproduced by a $19.95 CD player.



and on a nice system they are enhanced.


No, musical values are not enhanced by listening to a really good system. In general, people who are most concerned with musical values are capable of doing their musical work using fairly primitive tools by audiophile standards.

OTOH, what is enhanced by a really good system is the overall experience. It is easier for people who are not vitally involved with music to become more engaged with the whole musical performance if it is reproduced well, unless the quality of the reproduction becomes distracting.


When I look at a brochure for a new amp this is what I'ld like to see.

Musicality is a property of music, not audio gear.

The more I work with music and musicans, the more I realize this.

The OP asked "What part of sound is not measurable? "

My answer: For any point in time and space there is only ONE parameter to measure:

PRESSURE

The OP asked "What part of sound is not measurable? "

My answer: For any point in time and space there is only ONE parameter to measure:

PRESSURE
Yes, if all you want to know is pressure. There is also particle velocity, which is of course a vector. It is easily measured via a directional velocity transducer, such as a soundfield-type microphone.

- Terry

You probably only think that way because you haven't thought things out. Presumably you do care what the system sounds like and if you understood measures, you'd understand that there is a strong relationship between how a systems measures and how it sounds.

Hi Arny,

I think this comment points to the disconnect at hand.

If we take another step back from the problem, we realize that it IS possible to get at least a good indication of how a loudspeaker will sound if you took it outdoors, or even provide a means for a listener to listen to a speaker in isolation through convolution and headphones. The problem is that NO ONE listens to a loudspeaker or any other component of their system in isolation. This isn't news to anyone who has tried to measure a loudspeaker in a small room, but it is the crux of the issue.

We have lots of measurements of loudspeakers and we then have many who hear a speaker that looked to measure well and the sound they hear is terrible. The vast majority of those interested in hi-fidelity audio judge sound by the presentation of the system they hear, which as we all know, means the room acoustics, as well as placement of speakers and listener dictate how the loudspeaker's performance is translated to the listener. More often than not, the translation is not very good, and in plenty of cases preference comes down to what speaker does the least to excite offensive behavior of the room.

While it is much easier to generally define our subjective expectations in the great outdoors, real rooms suddenly require much more information to get even close to the same level of confidence. Likewise, the same aspects are not as important, as others that have no significance outdoors are highly important in our listening rooms. I know we can predict the sound in a room, I just don't think most grasp the exponential increase in complexity and number of variables a listening room adds to the problem.

In fact both questions are of interest, what you will buy and what you should buy. Ideally they would be the same.

That's amusing, and reminiscent of Emanuel Lasker's Machaïde (or Macheide), a being which has attained such intellect that it has lost its free will, since it can foresee that all courses of action other than the one it will pursue would turn out badly.

Yes, as there is no standardized reproduction environment for 2-channel audio, as there is for motion picture multi-channel sound.


And since different people prefer different direct/reverberent characteristics, I don't see that a standard is even a reasonable thing to suggest. That is, at least until capture and synthesis (i.e. studio recordings) have a standard, and I think that's counter to artistic intent.


But I think it can be generally agreed* that flat frequency response, both on and off axis, are minimally necessary for accurate reproduction. That is straightforward to measure and plot, and relatively easy to interpret visually.


It's a start.


Not for the most general meaning of the term. It can include many linear phenomena, such as frequency and phase distortion.

I try to keep words to a specific meaning. Keep distortion to refer to nonlinearities. Use frequency and phase shaping to describe frequency and phase shaping, don't call them distortion unless there is a nonlinear component.

Noise is the one ringer, I tend to lump it in with distortion since it is also a lossy process.

You are treating the recommendation problem as though it were the scientific problem of predicting what the listener will choose. I understand why a scientist would do this, but I think it misses the point that most engages us consumers. If I, as a layman, ask you, as an expert, to tell me what the best speaker is, I probably don't mean that I want you to predict what I will buy. Instead, I want you to tell me what's best -- what I should buy.

And, for that, I must know your preferences. This has nothing to do with "predicting", it has to do with knowing what YOU, personally, like.

While it is much easier to generally define our subjective expectations in the great outdoors, real rooms suddenly require much more information to get even close to the same level of confidence. Likewise, the same aspects are not as important, as others that have no significance outdoors are highly important in our listening rooms.
For those of you who are not aware, the great outdoors is nature's own anechoic chamber. And it can have better very low frequency performance than the best acoustic laboratories! :)

The OP asked "What part of sound is not measurable? "

My answer: For any point in time and space there is only ONE parameter to measure:

PRESSURE

Please allow me to point out the error a bit more strongly.

For any point in space, at any time, there are 4, yes, four measurements requred to fully characterize the soundfield.

Three volume-velocities and one pressure, to be specific.

And since different people prefer different direct/reverberent characteristics, I don't see that a standard is even a reasonable thing to suggest. That is, at least until capture and synthesis (i.e. studio recordings) have a standard, and I think that's counter to artistic intent.

It is for mixing rooms, which is where the sound for speaker reproduction is cast in stone (er, bits). Regardless of diverse studio or live venue recording techniques, the recorded sound is what is heard in the mixing room, through monitor speakers. And those mixing rooms have no standards.

I want to be glib and say it isn't—that you listen to the performance, and you decide it sounds sad to you. But of course it may well be that the performer is trying to convey "sadness," and we have conventions for doing that which you interpret as the performer intends you to. But there's no guarantee that you'll interpret something the way the performer intended it, nor would you be in any way "wrong" to interpret it differently. So to say that the emotions you feel as you listen to music are "transmitted" to you seems wrong to me.

But let's assume that you and the performer are on the same wavelength, and you get what he's driving at. How does he do it? Well, he controls aspects of the sound—loudness, tempo, timbre, etc. All of those things are definitely measurable. And therefore we can measure the extent to which an audio system preserves those elements so that you might interpret them as intended.

This kind of discussion is hard to describe because music is open to much interpretation:

watch this:
http://www.youtube.com/watch?v=5fybBwlxHNc

So here the intent is to make you jump or scare you. The highest level senders comprised of the directors and actor and the highest level reciever is you. The probably of "scare" transmission, lets guess at 50%. If I were to turn off the volume the scare transmission drops to 5%. If I were to raise the volume to peak at 110 dB then the scare transmission goes up to 60%.

To get to me the scare transmission has to undergo an infinite number of transforms. Each transform is orthogonal to each other. Each transform is a sender must be undone by a reciever to correctly transmit the signal. Each reciever/sender pair operates on the same principles. When done with 100% transmission I can model the reciever/sender pair as a wire.

Pretend the video was a DVD. There is a camera/television pair, an electrical A/D, D/A pair and an optical writer/reader pair. The transmission looks like this:

Director
Camera
A/D
Optical Writer
Optical Reader
D/A
Television
Me

A poor A/D; D/A conversion will result in a loss of transmission for the camera/television pair and will affect the director/me transmission.
In between each pairing is an infinitly complex layer of transforms coexisting on different planes. Our understanding of the principles behind cameras and televisions and A/D and D/A and optics is quite high, hence why it is easy for us to measure. Our understanding of the principles behind the human mind is not so good.

While it is much easier to generally define our subjective expectations in the great outdoors, real rooms suddenly require much more information to get even close to the same level of confidence. Likewise, the same aspects are not as important, as others that have no significance outdoors are highly important in our listening rooms.

For those of you who are not aware, the great outdoors is nature's own anechoic chamber. And it can have better very low frequency performance than the best acoustic laboratories! :)

Thank you Terry.

I chose the example of outdoors as this is a tangible environment available to most, vs. a theoretical anechoic chamber or one existing in only a few labs. The point is very important though... A good speaker as viewed in isolation is not the same as a speaker well suited for your room, assuming the room & locations even hold the potential for "good" results.

This is where things get tricky. I get a kick out of undefined mentions of "flat response." Flat with what scale? Zooming in or out on the vertical scale can make any real speaker's frequency response take on the shape of either the great plains or the rocky mountains using the exact same measurement. :rolleyes:

I would argue the concept of "flat enough" gets very vague when you have the much more complex situation of a small room. In real rooms the far off axis response variations and lobing can hugely affect the subjective sound quality of a speaker with even the best response imaginable in the forward 30 degrees of radiation. Within the confines of real, affordable devices we can build, there can be compromises in some of the most commonly examined measures which will allow for a better subjective result through significant improvement in other important characteristics... Characteristics which have little significance in mother nature's anechoic chamber.

So, talking about listening to the music is completely out of place in a discussion of high quality audio products.


That's where we disagree.

Pretend the video was a DVD. There is a camera/television pair, an electrical A/D, D/A pair and an optical writer/reader pair. The transmission looks like this:

Director
Camera
A/D
Optical Writer
Optical Reader
D/A
Television
Me

A poor A/D; D/A conversion will result in a loss of transmission for the camera/television pair and will affect the director/me transmission.
In between each pairing is an infinitly complex layer of transforms coexisting on different planes. Our understanding of the principles behind cameras and televisions and A/D and D/A and optics is quite high, hence why it is easy for us to measure. Our understanding of the principles behind the human mind is not so good.

You are correct that our understanding is not as good as in the video sense, yet we are also talking about a much wider bandwidth with much more complexity in propagation and interpretation by the listener.

Your example still allows for simple experimentation in starting with a case of good communication of these factors. Start inducing distortions or omissions and compare how significantly the communication is deteriorated. The same has already been done at great length for speech intelligibility with much learned from the process.

A poor A/D; D/A conversion will result in a loss of transmission for the camera/television pair and will affect the director/me transmission.
I presume by "loss of transmission," you do not mean "complete loss of transmission." Of course it can't scare you if you don't see/hear it.

But if that's not what you mean, then you are assuming your conclusion. Given that the D/A conversion happens at all, why should we think it would be scarier, or more likely to be scary, observed on a high-end system rather than through tinny computer speakers? I don't even think making it louder would have the impact you ascribe to it.

I don't buy the notion that highly accurate reproduction is necessary to appreciate an artist's intent. I can "get" Beethoven, and even someone's interpretation of Beethoven, on a boombox. It's not how I choose to listen most of the time, because I also enjoy highly accurate reproduction, and the sense of "being there." But to me, being an audiophile and being a music lover are two different things.

Thank you Terry.

I chose the example of outdoors as this is a tangible environment available to most, vs. a theoretical anechoic chamber or one existing in only a few labs. The point is very important though... A good speaker as viewed in isolation is not the same as a speaker well suited for your room, assuming the room & locations even hold the potential for "good" results.

I don't want to have to buy a different speaker for each listening room. I don't want to have to buy loudspeakers based on whether they are 'well suited to my room' except perhaps in terms of SPL, any more than I want to buy an amp or CD player 'well suited to my room'.

So to me a good loudspeaker is one that measures well (as per work relating measurements to listener reports of 'good sound' under well- controlled conditions), I can then EQ the sound or treat the room to capitalize on the performance of the loudspeaker.

It may be that I turn out to be the minority that actually prefers a sound *different* from the Toole/Olive paradigm, but I can only find that out by first knowing what sorts of speakers I am listening to...by virtue of their measured performance.

I don't want to have to buy a different speaker for each listening room. I don't want to have to buy loudspeakers based on whether they are 'well suited to my room' except perhaps in terms of SPL, any more than I want to buy an amp or CD player 'well suited to my room'.

So to me a good loudspeaker is one that measures well (as per work relating measurements to listener reports of 'good sound' under well- controlled conditions), I can then EQ the sound or treat the room to capitalize on the performance of the loudspeaker.

It may be that I turn out to be the minority that actually prefers a sound *different* from the Toole/Olive paradigm, but I can only find that out by first knowing what sorts of speakers I am listening to...by virtue of their measured performance.

I am not suggesting that you have to buy a different speaker for every room, but rather that the breakdown and skepticism of measurements by enthusiasts stems from the reality that a small set of measurements which give a decent indication of a loudspeaker outdoors, is grossly insufficient in a real room.

I most certainly believe measurements can tell us a great deal or nearly everything about the sound of a speaker. I don't believe 4-5 2D measurements will ever be sufficient to set our expectations properly in real listening rooms. Most talk of loudspeaker measurements I see don't tell us what we really need to know, and what we need to know isn't yet as simple to examine and decipher as those that apply to anechoic situations.

It is for mixing rooms, which is where the sound for speaker reproduction is cast in stone (er, bits). Regardless of diverse studio or live venue recording techniques, the recorded sound is what is heard in the mixing room, through monitor speakers. And those mixing rooms have no standards.

You're talking about mixing rooms. I agree that we need standards for mixing rooms. I've been cheering Tom Holman on for a long time on that one.

Acoustics, levels, ...

And multichannel means that a lot of older rooms are just WRONG. You don't do multichannel mixing in LEDE. No no no...

But watch the flames on that one.

I get a kick out of undefined mentions of "flat response." Flat with what scale? Zooming in or out on the vertical scale can make any real speaker's frequency response take on the shape of either the great plains or the rocky mountains using the exact same measurement. :rolleyes:

And what about "smoothing" the curve? (mutter)

And what about "smoothing" the curve? (mutter)
Standardization is definitely in order here. 3rd octave smoothing is probably a good and practical first step. EBU 3276 specifies that the ISO standard 1/3 octave discrete bands be used, but that is way too coarse for many purposes. A sliding filter window shows a great deal more detail.

We need an exact spec on how to perform such smoothing. I have never seen one in print. And what length of FFT do we use to calculate the frequency response? What about the actual length of the impulse response, vs. how many zeros are padded at the end to give it the required length for the FFT? And exactly what type of windowing do we apply to the impulse response?:eek:

Personally, I'd rather have direct access to the measured impulse response, and do my own frequency and time analysis. But this completely raw data would be totally useless to anyone trying to directly evaluate and compare performance.

Standardization is definitely in order here. 3rd octave smoothing is probably a good and practical first step. EBU 3276 specifies that the ISO standard 1/3 octave discrete bands be used, but that is way too coarse for many purposes. A sliding filter window shows a great deal more detail.

We need an exact spec on how to perform such smoothing. I have never seen one in print. And what length of FFT do we use to calculate the frequency response? What about the actual length of the impulse response, vs. how many zeros are padded at the end to give it the required length for the FFT? And exactly what type of windowing do we apply to the impulse response?:eek:

Personally, I'd rather have direct access to the measured impulse response, and do my own frequency and time analysis. But this completely raw data would be totally useless to anyone trying to directly evaluate and compare performance.

Terry, I assume you are familiar with the CLF group/format (http://www.clfgroup.org/) which was started by most of the acoustic modeling software companies apart from EASE? I'm a big fan of their free viewing tool as a compliment to more detailed/typical data. Interestingly, most all of the data is gathered from impulses. I oversaw getting some speakers from the company I slaved for in a past life to be the first published in the format by a 3rd party testing. Pat Brown of SynAudCon/ETC, Inc. and Ron Sauro of NWAA Labs both provide validated, 3rd party testing for reasonable dollar amounts giving real 10-0.5 degree increment spherical, not just polar (H/V), data.

The perspective of the data at the moment is geared more toward the larger pro audio venue, but is still of interest, and is rather flexible in viewing. Again, understanding what it might be telling you is still a hurdle. By the end of the year I'll have my own designs tested in this manner. I have heard from Pat & Ron that a few hi-end audio manufacturers who have been asked to submit their speakers were rather surprised with what they saw in their shipping products:rolleyes:. Of course there are a very few manufacturers who have such in-house capabilities.

We should keep some perspective that home/consumer audio is still rather naive to the 3D function of our loudspeakers, even while being so particular about other laudable aspects.

Yes, Mark. I too am a big fan of CLF. One can do CLF 1 format, which has 10 degree spherical resolution at octave frequencies, or CLF 2 format, with 5 degrees spherical resolution at 1/3 octave frequencies (if I remember right).

I'd like to see an EASE to CLF translator which does not require purchase of EASE. Pro loudspeaker manufacturers seem to publish the EASE formats more than CLF.

And of course I'd like to see every speaker manufacturer in the world, pro and consumer, measure and publish their CLF data! At $1350 from NWAA labs, this is not prohibitively expensive.

I get a kick out of undefined mentions of "flat response." Flat with what scale? Zooming in or out on the vertical scale can make any real speaker's frequency response take on the shape of either the great plains or the rocky mountains using the exact same measurement. :rolleyes:

And what about "smoothing" the curve? (mutter)

Part of my point was that in the presentation of a graph, there are TWO means to smooth... You can process/smooth the actual data, or smooth the display with scaling and thick markers. With the latter they get to still claim the measurement is of higher resolution. ;)

In the loudspeaker measurements I have posted on my own forum or here, I prefer to post multiple resolutions of the same data to burn different perspectives in the mind before comparing to a curve smoothed at 1/3rd octave on a 120dB vertical scale. :rolleyes:

different people prefer different direct/reverberent characteristics

Yes, and people also learn to appreciate improvements and change their opinion over time. I often characterize a well-treated listening room as an acquired taste. Some people say they like their small room's ambience because it retains "liveness," and if you treat their room some initially feel it has lost something. But most soon realize that their room was adding a bad quality to everything they listened to, and once they hear music without all the early reflections and comb filtering they realize the sound is actually much larger without the added ambience.

--Ethan

If I want larger sound I'll put on some music sung by chubby chicks.

It's rainin' men! Hallelujah! ;)

Yes, and people also learn to appreciate improvements and change their opinion over time.
Absolutely. I'm willing to take my preferences into account, but I'm not prepared to cater to them. I want to refine my understanding of high fidelity sound.

Then you're looking in the wrong place. Try a fMRI.

Hey, that's my line. ;)

I need to find some time to get back to AVS, saw this thread start up but haven't had time to follow it, interesting discussion!

I see no evidence that you have some unique qualifactions to speak for most audiophiles. You don't know what most people care about. You can only speak for yourself. You shouldn't blow yourself up like this in public - its embarassing for the rest of us.

I see no evidence that it's a requirement for anyone to justify their "qualifications" to a bunch of randoms on a chat room. You could be a little more diplomatic and assume that the man has a little knowledge in the subject--but I'd then figure that you'll look for things that may indicate he doesn't.

People who listen primarily to the music could in general care less what the recording sounds like.

So, talking about listening to the music is completely out of place in a discussion of high quality audio products.

Fact is that all of the above are well-reproduced by a $19.95 CD player.

No, musical values are not enhanced by listening to a really good system.

I think you just need to put in your signature "I just don't get it."

Everything you mentioned above is VITALLY important to listeners of music. The role of good gear is to enhance the experience, and the music is the point. You either contradict yourself or are detached from this hobby so much you might as well not even comment.

If I were to make fun of a pure objectivist, I would have typed everything above you did. Thanks for taking care of my light work.

Kinda off topic a bit but do you always want "accurate" sound?

Accuracy is a very desirable and worthwhile starting point. Sure, you can always add whatever flavoring you like but I would see anything such that cannot be controlled or defeated as sloppy design.

I might like some tinting or lighting or other effects when I do photography, but I would regard any camera, monitor, photo processing software, et al, that arbitrarily and universally adds these modifications as faulty.

Accuracy is a very desirable and worthwhile starting point. Sure, you can always add whatever flavoring you like but I would see anything such that cannot be controlled or defeated as sloppy design.

I might like some tinting or lighting or other effects when I do photography, but I would regard any camera, monitor, photo processing software, et al, that arbitrarily and universally adds these modifications as faulty.

Here we go, a reasonable post. Took how long?