Speaker Testing: why mono is better

[Spocko]

That is only one half of the evaluation. Let us say you have two speakers that don't sound terrible in mono. What can you say about how they sound while used in stereo as they typically are? Will both sound the same or very different in a way that cannot be inferred from a mono evaluation. This is the crux of the question.

So, it depends on the goal. Are you trying to detect "anomalies" in a speaker or are you trying to objectively measure/quantify the audible qualities of a device? This is why I keep saying that you can take a review here as a spectator sport of relative grading (former goal) or as a buying guide based on measurements (latter goal). Two entirely different goals that cannot be conflated with each other or assume one necessarily implies the other.

Moreover, I am not even sure the bolded portion above is necessarily true. There is no binary of "terrible" and "not terrible". There are various grades of deviations in multiple dimensions. So, the "stereo" effect may mask some degree of "terrible" but not some other degree of terrible. Unless you quantify and establish a threshold for the latter, you cannot assume any degree of terrible isn't masked by stereo.

(1) One can say measuring stereo is difficult or that there is no standardized way to set up and test a stereo speaker. That may be so but it doesn't imply the testing in stereo isn't necessary or useful.
(2) One can say that two speakers that measure the same in mono (which is entirely theoretical since no two speakers will ever measure the same) won't sound any different in stereo. While, there may be some justification for this in electronics, it becomes a lot more tenuous in evaluating speakers since there is no such thing as a perfect/transparent speaker.

So the whole argument for the case of mono testing boils down a very flawed and artificial constraint on the speaker behavior.

A speaker so terrible that stereo won't save it (and so stereo isn't necessary) but not so terrible that only mono measurement can expose it (and so mono is necessary).

At best, that seems like a self-serving rationale for (1) above.

I think you raised a very good point about stereo testing: there is no standardized way to set up a test for stereo listening. This leads to the conclusion that current reviews in stereo are at best a crapshoot - some may be good some may be inaccurate. This reduces the value of stereo reviews - although not eliminating it completely, you are now left wondering how much of the review is about the room interaction and how much is the pure speaker? I just believe the heuristic for speaker performance is much more predictive with mono testing than stereo testing. I'm not saying that stereo does not offer unique performance measures that should be investigated but I do not believe it would contradict the conclusions of a mono review.

 

[Vasr]

I think you raised a very good point about stereo testing: there is no standardized way to set up a test for stereo listening. This leads to the conclusion that current reviews in stereo are at best a crapshoot - some may be good some may be inaccurate. This reduces the value of stereo reviews - although not eliminating it completely, you are now left wondering how much of the review is about the room interaction and how much is the pure speaker? I just believe the heuristic for speaker performance is much more predictive with mono testing than stereo testing. I'm not saying that stereo does not offer unique performance measures that should be investigated but I do not believe it would contradict the conclusions of a mono review.

Just saying the above is purely ideological faith in a "testing method" not a science-based approach.

You cannot justify a potentially flawed or insufficient measurement by pointing out "bigger" flaws in some other measurements. Both can be flawed enough to not get at the truth (i.e., model reality sufficiently). Just pointing out a logical fallacy.

Imagine saying at some point in history we believe the earth to be flat by looking at the 1 foot level here because the measurements to measure curvature of earth are fraught with inaccuracies or can be affected by other factors.

Or that there are no other living organisms in the universe because any evidence to the contrary will be outdated by the time it gets here, etc....

 

[amirm]

That is only one half of the evaluation. Let us say you have two speakers that don't sound terrible in mono. What can you say about how they sound while used in stereo as they typically are? Will both sound the same or very different in a way that cannot be inferred from a mono evaluation. This is the crux of the question.

And the question that was answers in research and in the video. Listeners get so lost in the content in stereo that they can't do their job properly in evaluating said speaker.

Stereo selling is why so many lousy designs sell. Nothing about stereo fixes the flaws in speakers.

If you feel otherwise, then you need to present research to back that point of view. Not, "doesn't this sound right" argument. The whole research is about how our intuition about testing in stereo is wrong. So you can't appeal to gut feeling.

 

[amirm]

Just saying the above is purely ideological faith in a "testing method" not a science-based approach.

Did you watch the video? "Science based" research from top luminaries in speaker and sound reproduction was presented in this video. There is no faith here. It doesn't get more scientific than what was presented. Not agreeing just because is definition of faith based belief.

 

[KSTR]

The excess phase of a multiway speaker usually refers to the excess above and above a minimum phase response, typically caused by the filters used to implement the crossover. Assuming a pair of speakers built to a reasonable tolerance, what is the impact on imaging (beyond off axis response)?

I'm not aware of any rigorous studies on this but what I have found (at my then day job and later at home) is that a linear phase speaker (absent excess phase) tends to have a clearer imaging and more "ambience resolution" in stereo than one would expect from the audible improvement when assessed as single speaker where the effect is mostly reduced to "transient compactness" and timbre shifts. These were some experiments (blinded A/B) within the constraints of a company, often with tight schedules...

Reverb tail and phantom source rendering etc just seem to be emphasized by linear phase. Working hypothesis is that the group delay vs frequency of minphase XO'd speakers partly falls in a range that is equivalent to time-of-flight difference between ears, per speaker. This might interfere with localization cues based on interchannel time differences encoded in the source signal (A/B stereo, vs X/Y where only magnitude differences create the localization cues), making our brains work harder to follow whats going on. A factor here is also phase coherency between ways, the better the match the more making the total response minimum phase pays off.

-------------------

In closing, I will say that I fully agree to what @amirm does in his speakers test: first things first. While one can fix frequency response of magnitude and phase (with proper DRC, for example) you cannot fix directivity and distortion issues and that's why these tests and measurements are very valuable. The perceptual rating from mono listening just does not cover the whole story, though. IHMO, YMMV.

 

[Racheski]

Thank you Amir for this informative video - I really needed this before I went to ABT 2 years ago and spent $$$ on B&W speakers.

 

[Spocko]

I'm not aware of any rigorous studies on this but what I have found (at my then day job and later at home) is that a linear phase speaker (absent excess phase) tends to have a clearer imaging and more "ambience resolution" in stereo than one would expect from the audible improvement when assessed as single speaker where the effect is mostly reduced to "transient compactness" and timbre shifts. These were some experiments (blinded A/B) within the constraints of a company, often with tight schedules...

Reverb tail and phantom source rendering etc just seem to be emphasized by linear phase. Working hypothesis is that the group delay vs frequency of minphase XO'd speakers partly falls in a range that is equivalent to time-of-flight difference between ears, per speaker. This might interfere with localization cues based on interchannel time differences encoded in the source signal (A/B stereo, vs X/Y where only magnitude differences create the localization cues), making our brains work harder to follow whats going on. A factor here is also phase coherency between ways, the better the match the more making the total response minimum phase pays off.

-------------------

In closing, I will say that I fully agree to what @amirm does in his speakers test: first things first. While one can fix frequency response of magnitude and phase (with proper DRC, for example) you cannot fix directivity and distortion issues and that's why these tests and measurements are very valuable. The perceptual rating from mono listening just does not cover the whole story, though. IHMO, YMMV.

Your last statement is being pursued by many as more important so what's lost in this debate - mono listening is clearly valuable and serves a purpose but does not "cover the whole story," so what we should be discussing is not the value of mono listening (clearly it works for its intended purpose) but rather what else could a reviewer reasonably do to help tell the whole story if mono alone is not enough? Do we agree that all stereo reviews must be conducted in a parallelogram 12 x 10 room with 8 feet ceiling and the same room treatment used in the exact arrangement collectively agreed to by acousticians as best for stereo listening? Or possibly three different room categories: large room, medium room and small room?

 

[Sancus]

My experience has been rather different. Variations of several dB when measured on-axis, mostly around crossover frequencies, but also elsewhere. Very good loudspeakers manage to be within 1dB from 100Hz to 15kHz, but many more are likely to have variances of several dBs.

If current loudspeakers are normally within 0.5dB variation, then indeed, it's no longer an issue, but how would we ever know unless we see the graphs?

Yeah it's not true that speakers are generally matched to 0.5dB. Here is an example of the JBL 306ps which are 2dB off. The better studio monitors are matched to within 0.5dB or less(0.26dB for the Neumann KH80) relative to every single speaker sold, not just by pair, which is why they have no reason to sell them in pairs. That also better matches their use -- plenty of studios need multi-channel setups and have no use for pair matching.

 

[krabapple]

I've read the original Toole/Olive work and also read their patient responses to these objections over the years. And yet here we are, people recycling the same objections.

 

[Daverz]

Imagine saying at some point in history we believe the earth to be flat by looking at the 1 foot level here because the measurements to measure curvature of earth are fraught with inaccuracies or can be affected by other factors.

I'd say in that case you'd be justified in tentatively holding to the simpler hypothesis. It will work OK locally.

Luckily, all you need to see the curvature of the Earth is a long enough and calm enough body of water (though it won't convince a flat Earther).

https://en.wikipedia.org/wiki/Bedford_Level_experiment

 

[JimWeir]

This topic of "why Amir tests speakers in mono" keeps coming up. I must have explained that a hundred times in text. Most of you probably know why. But I thought I do a video that covers the research and explains it all. Here it is:

References:
https://www.aes.org/e-lib/online/browse.cfm?elib=11740
Subjective Measurements of Loudspeakers: A Comparison of Stereo and Mono Listening

https://www.aes.org/e-lib/browse.cfm?elib=14622
Comparison of Loudspeaker-Room Equalization Preference for Multichannel, Stereo, and Mono Reproductions: Are Listeners More Discriminating in Mono?

https://www.aes.org/e-lib/browse.cfm?elib=8338
A New Laboratory for Evaluating Multichannel Audio Components and Systems

This is so true. When voicing speakers, listening for directional anomalies, and certain looking for dynamic variations in tones or distortion mono rules. What the AES papers don’t really address is why. And that has to do with the illusion of stereo reproduction, the cool trick that brought us into the hobby to begin with. With stereo, you have two somewhat correlated pressure variations at the ears, delivered by two or more loudspeakers spaced apart. (Yes multichannel is “stereo”, Greek for “solid” as in geometry, a cube or a pyramid for example, a three dimensional object.) the illusion, as mentioned above, is that multidimensional rendering of a performance. When presented with those differing aural inputs the brain engages more regions of the brain and directs its resources to create that imagined reality.

In mono, one speaker or two, the brain can focus on the content. Does that trombone sound like my memory says it should? I very much separate my analytical content and processes from my pleasure seeking listening sessions, in various stereo modes.
Oh, I’ve you have two similar sounding loudspeakers in mono, their presentation with stereo signals will sound good two!
Good for DYI loudspeaker developers, you only need one prototype.

 

[pozz]

I'm not aware of any rigorous studies on this but what I have found (at my then day job and later at home) is that a linear phase speaker (absent excess phase) tends to have a clearer imaging and more "ambience resolution" in stereo than one would expect from the audible improvement when assessed as single speaker where the effect is mostly reduced to "transient compactness" and timbre shifts. These were some experiments (blinded A/B) within the constraints of a company, often with tight schedules...

Reverb tail and phantom source rendering etc just seem to be emphasized by linear phase. Working hypothesis is that the group delay vs frequency of minphase XO'd speakers partly falls in a range that is equivalent to time-of-flight difference between ears, per speaker. This might interfere with localization cues based on interchannel time differences encoded in the source signal (A/B stereo, vs X/Y where only magnitude differences create the localization cues), making our brains work harder to follow whats going on. A factor here is also phase coherency between ways, the better the match the more making the total response minimum phase pays off.

-------------------

In closing, I will say that I fully agree to what @amirm does in his speakers test: first things first. While one can fix frequency response of magnitude and phase (with proper DRC, for example) you cannot fix directivity and distortion issues and that's why these tests and measurements are very valuable. The perceptual rating from mono listening just does not cover the whole story, though. IHMO, YMMV.

When you say a linear phase speaker, what does that mean? If you're talking about multiway speakers with noncoincident drivers (like your designs, as far as I know), phase can only ever be linear relative to one position with a speaker. A speaker designed for linear response on axis will not be linear elsewhere. If I'm wrong please correct me.

From what I know, under anechoic conditions, 60 degrees of phase shift within one critical band is considered audible. In rooms, I don't think your working hypothesis holds when you start adding the effects of reflections. The phase error in the loudspeaker would have to be extremely gross. John Vanderkooy did a study that showed little audibility for shifts adding up to several cycles. If you also account for the fusion interval for localization, which is measured in tens of milliseconds, and then phase locking for bass taking upwards of 200ms, phase doesn't really enter the picture other than when comparing an ideal situation to a typically measured one.

Regarding the topic at hand generally: a paper by Earl Vickers: https://www.sfxmachine.com/docs/FixingThePhantomCenter.pdf

 

[Ismapics]

This is so true. When voicing speakers, listening for directional anomalies, and certain looking for dynamic variations in tones or distortion mono rules. What the AES papers don’t really address is why. And that has to do with the illusion of stereo reproduction, the cool trick that brought us into the hobby to begin with. With stereo, you have two somewhat correlated pressure variations at the ears, delivered by two or more loudspeakers spaced apart. (Yes multichannel is “stereo”, Greek for “solid” as in geometry, a cube or a pyramid for example, a three dimensional object.) the illusion, as mentioned above, is that multidimensional rendering of a performance. When presented with those differing aural inputs the brain engages more regions of the brain and directs its resources to create that imagined reality.

In mono, one speaker or two, the brain can focus on the content. Does that trombone sound like my memory says it should? I very much separate my analytical content and processes from my pleasure seeking listening sessions, in various stereo modes.
Oh, I’ve you have two similar sounding loudspeakers in mono, their presentation with stereo signals will sound good two!
Good for DYI loudspeaker developers, you only need one prototype.

Yes it sound like what the loudness button use to be. I have Linn front stage, polk surrounds and Velodyne sub. Cheers

 

[Vasr]

Did you watch the video? "Science based" research from top luminaries in speaker and sound reproduction was presented in this video. There is no faith here. It doesn't get more scientific than what was presented. Not agreeing just because is definition of faith based belief.

First, I offered no belief (pro or con) to have faith in. It is part of science to not agree with methodology or its shortcomings. I have taken no position on whether stereo measurement would or would not offer something if there is a way to measure it. So the above statements make zero sense.

I was referring to the quoted post which said

I'm not saying that stereo does not offer unique performance measures that should be investigated but I do not believe it would contradict the conclusions of a mono review.

That belief is of faith not of evidence. Primarily because it assumes without evidence that the perceived performance of a stereo set up will not mask some of the conclusions of a mono review. This is not the same thing as saying that the speaker on its own will behave better in the same measurements in a stereo setup for some magical reason. This is part of the confusion with monoists (if I may call them that).

There is a logical fallacy continuously being used here to justify monoism that percolates into the video as well. That is well demonstrated in this post:

And the question that was answers in research and in the video. Listeners get so lost in the content in stereo that they can't do their job properly in evaluating said speaker.

Stereo selling is why so many lousy designs sell. Nothing about stereo fixes the flaws in speakers.

If you feel otherwise, then you need to present research to back that point of view. Not, "doesn't this sound right" argument. The whole research is about how our intuition about testing in stereo is wrong. So you can't appeal to gut feeling.

The first statement is like saying that people get so lost in the experience of eating an entree that they can't do their job properly in evaluating the bitterness of the truffles in it. We can correctly evaluate the level of bitterness of truffles if we taste truffles on its own. These are correct statements but is also an absurd view to hold.

It eventually boils down to what is it you are trying to do. Document the bitterness of truffles on some scale or try to objectively measure somehow whether one entree with same ingredients tastes better than another using a model for the latter. All I am saying is that that latter model does not exist (yet) for speakers. The fact that we have a model for measuring truffle bitterness doesn't imply we have the required model for measuring an entree. They are not mutually exclusive either. And yet people seem to persist in doing so as a matter of ideology than science.

This is not a debate of results but of methodology (application of science and interpretation) if one can understand the difference. Logic is the arbiter of this not more studies.

 

[JimWeir]

Thanks Amir,

that mono speakers are more sensitive for evaluations are quite clear, but did you bring up the issues with the flaws of stereo (comb filtering)? Which means that a mono speaker sounds different from a mono source played by a stereo setup, regardless of the speaker is perfect or not.

Yes, spread speakers playing is a different evaluation. When comparing two speakers as a matched pair place the speakers close together and evaluate their sound with mono signal individually.

 

[restorer-john]

There is no way to determine slight channel mismatch in stereo listening.

Absolutely there is. I do it all the time. The entire chain including the speakers can be assessed for channel balance in a second or two. Note. You need to consider "stereo" vs 2 channel. Two speakers and two amplifiers does not equal stereo. The same content (summed mono) in both speakers is way more useful than either 1 single speaker or stereo content in two speakers.

Not only can "slight channel mismatch" be assessed, crossover anomalies between samples can be identified and potential outliers, or (as you have found out on several occasions already) faulty/damaged speakers. A single speaker test/listen/review fails to identify any of those issues as there is no second sample to test and no second reference to compare with. It is like picking a single item of a production line and declaring the entire shipment is either faulty, or good to go, based on one sample.

The purchase case, the usage case and the content delivered, all demand two speakers, so to ignore, dismiss and actively deride basically the entire reason for modern high fidelity as we know it, is hilarious. But I do acknowledge, testing, shipping and reviewing a single speaker is logistically easier.

Nobody is saying single speaker auditions are not useful, but they are not the be all, and end all, of speaker reviewing. It is certainly not the only correct way to review what is essentially only half a product and to suggest so, is somewhat arrogant IMO. You test both channels of every D/A converter, power amplifer and preamp- nobody says you should only test the left channel and listen to that in isolation, do they? Why? Because the two channels a) affect one another and b) are designed to be listened to together.

When anything is effectively reviewed, the typical use case must be auditioned and considered. That use case is stereo content, reproduced on a pair of speakers.

 

[whazzup]

Absolutely there is. I do it all the time. The entire chain including the speakers can be assessed for channel balance in a second or two. Note. You need to consider "stereo" vs 2 channel. Two speakers and two amplifiers does not equal stereo. The same content (summed mono) in both speakers is way more useful than either 1 single speaker or stereo content in two speakers.

Not only can "slight channel mismatch" be assessed, crossover anomalies between samples can be identified and potential outliers, or (as you have found out on several occasions already) faulty/damaged speakers. A single speaker test/listen/review fails to identify any of those issues as there is no second sample to test and no second reference to compare with. It is like picking a single item of a production line and declaring the entire shipment is either faulty, or good to go, based on one sample.

The purchase case, the usage case and the content delivered, all demand two speakers, so to ignore, dismiss and actively deride basically the entire reason for modern high fidelity as we know it, is hilarious. But I do acknowledge, testing, shipping and reviewing a single speaker is logistically easier.

Nobody is saying single speaker auditions are not useful, but they are not the be all, and end all, of speaker reviewing. It is certainly not the only correct way to review what is essentially only half a product and to suggest so, is somewhat arrogant IMO. You test both channels of every D/A converter, power amplifer and preamp- nobody says you should only test the left channel and listen to that in isolation, do they? Why? Because the two channels a) affect one another and b) are designed to be listened to together.

When anything is effectively reviewed, the typical use case must be auditioned and considered. That use case is stereo content, reproduced on a pair of speakers.

But your cited examples for having stereo, is mainly to have a 2nd unit to compare against the 1st.
Not trying to stir things, but it's still unclear what is that fundamental test case that requires a stereo setup.

For example, if someone designs speakers, he makes a prototype and listens to it, then makes a 2nd prototype (assuming construction quality is similar) and listens to both, what is the quality that he finds in stereo listening that will cause him to re-design / iterate on his design of the speaker unit? That he didn't notice in mono listening? Soundstage is 'not wide' enough? But wouldn't soundstage be something apparent to the designer even in mono?

Or to put in a different way, is it a realistic scenario for someone to be able to, with experience, be so familiar with the differences between mono / stereo listening, that he can develop enough shorthand cues to identify all issues in mono (and thus make things faster)? He's not trying to create music with different left / right panning.

To me, it's somewhat like people arguing that 'music' is not 'just test tones' and berate people for not using music to test...

 

[Pennyless Audiophile]

First post, so feel free to ignore me...

While I understand Amir's point and I am far from rejecting his explanation, I don't see how this is of any relevance at all but for logistics.
We always listen stereo and the masking of defects is a good thing: you may potentially have a subjectively good sounding speaker with inferior or flawed specs, so potentially cheaper.

 

[whazzup]

First post, so feel free to ignore me...

While I understand Amir's point and I am far from rejecting his explanation, I don't see how this is of any relevance at all but for logistics.
We always listen stereo and the masking of defects is a good thing: you may potentially have a subjectively good sounding speaker with inferior or flawed specs, so potentially cheaper.

Except for the flip side of the coin, where you potentially have a subjectively good sounding pair of speakers with inferior / flawed specs, but you're charged an arm and leg for that.

 

[Pennyless Audiophile]

Except for the flip side of the coin, where you potentially have a subjectively good sounding pair of speakers with inferior / flawed specs, but you're charged an arm and leg for that.

True! I suppose that some common sense may help here.