Lack of high-end speaker reviews

[Blumlein 88]

Here is another thing to think about: assessing subjectively factors like soundstage is extremely hard and effectively impossible to do reliably. I always, always ignore such subjective comments. These people don't even have a reference track they can use to determine such let alone have the ability. Indeed, give an expensive speaker to a reviewer and they seemingly always have great soundstaging.

There was the chink, chink, chinking track that could be used. It never caught on. It seemed to have some utility when I've used it. There were recordings where people stood at different distances which could have been used. But you are right of course there has been no standard, reference or any attempt beyond "it has great soundstaging". I do remember a few reviews where you get the comment soundstaging is not a strength of this speaker.

 

[Blumlein 88]

With respect, you are being deliberatively combative and you know it.

You know full well how a loudpseaker 'system'* is the only way to convey spatial details like depth, height and placement in a 2 channel stereophonic recording. A single loudspeaker has zero opportunity to do so.

*more than one speaker.

Okay, so, if testing in mono and stereo has shown speakers end up ranked in the same order, which from what is known they usually do, and if mono is more discriminating as in showing larger differences in listener preference, then how is listening in stereo a better test just because they get used in stereo? It would rarely change the ranking. If you use the results to find what speaker properties are most often desired, then what is gained by stereo testing? Unless the stereo test were more discerning or some property is desired for stereo that is different in kind or level in mono then why is stereo testing helpful? If speakers scoring highly in mono also do in stereo what is being missed by mono testing?

 

[Blumlein 88]

The ESL63 suffers from trying to do too much. There’s not much purpose to a point source below the Schroeder frequency, and it beams a lot in the upper frequencies despite the delay line. Genius design, ahead of its time, but flawed?

I wouldn't call it flawed, and I doubt Peter Walker considered it perfect. He did what he could with the technology available to make an exceptional commercially viable speaker. One of his ideas never made was to have a long single ESL panel across the end of the room, and to feed each channel into each end with delay lines between segments horizontally. Now you could have multiple panels with their own amp and use DSP to dynamically create a soundfield.

I've wished Quad rather than copying the old design, would also use that idea for an improved modern version. Something with more segments to better approximate a quasi-point source that would beam less or it probably is even possible to contour the directivity with the technology available now.

 

[Newman]

Maybe I should refresh my memory. What I remember is that all speakers tested with results in a narrower range in stereo than in mono. In other words the difference in best and worst was diminished. Generally the relative ranking stayed the same. I do seem to recall the Quads switched places with one speaker scoring relatively better in stereo. I love panel speakers and specifically the ESL 63s. I do wish more work was done to investigate how they are so well enjoyed yet test poorly. A case of if you like it don't worry about the test.

I think I put those charts in this thread post #247 yesterday? The relative ranking did stay the same. I think you are interpreting it correctly.

Of course, don't be surprised if someone blasts me for quoting and misusing "ancient" publications, and accuses me of baiting. (This seems to only happen when people don't like the results, ie doesn't match their sighted listening preferences and purchases.)

 

[Blumlein 88]

I think I put those charts in this thread post #247 yesterday? The relative ranking did stay the same. I think you are interpreting it correctly.

Of course, don't be surprised if someone blasts me for quoting and misusing "ancient" publications, and accuses me of baiting. (This seems to only happen when people don't like the results, ie doesn't match their sighted listening preferences and purchases.)

Thanks I missed that one somehow.

Maybe the take home message is the more speakers the better the score. With enough everything is good. So we all need Atmos. Dolby would be happy. The improvement by the Quad in stereo was much more than the others. I cannot remember if other lower scoring mono speakers are shown in stereo tests and if they improved by similar amounts or the Quad is an outlier.

 

[rynberg]

Stereophonic recordings are not monophonic recordings. They are not 2 channel monophonic recordings.

Define a "stereo" recording. What makes a "stereo" recording not a 2-channel monophonic recording?

The' images' produced are NOT wholly and solely determined by mere amplitude differences. Anyone with half a brain knows that.

Feel free to explain all the factors that create an image in the recording. Should be easy, it will only take half your brain.

Virtually all modern recordings are compiled with several to dozens of mono sources, primarily set in the soundstage by amplitude differences in the two channels. A modern recording creates a completely synthetic soundstage. Most vocals are set dead center, ie, set to the same amplitude in both channels, ie dual mono. Same with the bass. Mixers can then further play with amplitude, and then add EQ and some effects to create the effect of distance for particular source elements, if warranted. A lot of modern recordings don't really go to that level of effort. So, the primary soundstage in most recordings is PRIMARILY determined by amplitude differences between the two channels.

You would have more of a point if you were talking about stereo-miking at distance in a concert hall or something but that is a very small minority of modern recordings.

 

[MattHooper]

Virtually all modern recordings are compiled with several to dozens of mono sources, primarily set in the soundstage by amplitude differences in the two channels. A modern recording creates a completely synthetic soundstage.

Remember that while many sources are recorded mono, once you have added stereo reverb (or similar effect/processing) to that source, you have created a stereo sound.
Likewise, plenty of electronic instruments are stereo.

Of course the intricacies all depend on how anything ends up being mixed, but the point is that it's too simple to think that a sound captured with a mono mic stays mono.
I add stereo effects and reverb to what were originally mono sources, thereby creating stereo files, all the time.

 

[Duke]

Okay, so, if testing in mono and stereo has shown speakers end up ranked in the same order, which from what is known they usually do...

I think I put those charts in this thread post #247 yesterday? The relative ranking did stay the same.

The sound quality rankings stayed the same.

The spatial quality ranking changed in going from mono to stereo (E moves into first place in spatial quality in stereo, though the margin may be small enough to be considered insignificant). Note that the "sound quality" scoring does not include the "spatial quality" scoring; those terms are applied to two separate sets of characteristics.

The degree of differentiation between the speaker rankings looks like it is consistently decreased in stereo, which would make stereo evaluation less valuable as a comparison technique.

Imo this three-speaker sample is far too small for those of us without access to a large body of raw data to draw reliable conclusions from. I believe @Floyd Toole when he says that the sound quality rankings never change when going from mono to stereo. It is not clear to me that he has said the same thing about the spatial quality rankings. Has he?

 

[youngho]

From Toole: "Choisel and Wickelmaier (2007) found that listeners comparing a discrete center channel with a phantom center image generated
by a stereo pair in a normal room consistently rated the phantom image higher in perceptual dimensions of width, elevation, spaciousness, envelopment, and naturalness." I have some additional material regarding distance perception and a quote from JJ that I'll post later.

I hate when people vaguely refer to things based on their faulty memory without providing any usable references or links.

From @j_j at https://www.aes-media.org/sections/pnw/ppt/jj/soundfields_vs_human_hearing_edited.ppt:
“For center images, the two signals from the loudspeakers conflict very badly. The interference creates frequency shaping that is the inverse of first-arrival distance cues and that mimics the effects of positive elevation cues.“

Also from @j_j but at https://www.aes-media.org/sections/pnw/ppt/jj/auditory_mechanisms_01_28_21.pptx:
Distance perception: signal to reverb ratio, decorrelation of leading edge across frequency (more=distance)

From Risoud at https://www.sciencedirect.com/science/article/pii/S187972961830067X:
“Human subjects were shown to be able to be able to determine monaurally the vertical localization of high but not low-frequency sounds, probably due to the small size of the pinna, which allows it to interact only with short-wavelength sounds [38]. Sounds can be accurately located vertically only if they are complex; they include > 7000 Hz components; the hearer’s pinna is present [39]” and “Determining the distance of a sound source mainly depends on monaural cues, and is much easier for familiar sounds [41].Generally speaking, close distances tend to be overestimated and long distances underestimated…The direct-to-reverberant energy ratio [19] is the first distance cue….Initial time delay gap (ITDG) is the time gap between the arrival of the direct sound wave and the first strong reflection…Level is also a distance cue, distant sources giving rise to lower perceived level... Spectrum is another distance cue, high frequencies being more quickly muffled by the air: air absorption coefficient is higher the higher the sound frequency”

From Blauert at https://direct.mit.edu/books/oa-monograph/4885/Spatial-HearingThe-Psychophysics-of-Human-Sound, look at Chapter 4, page 360 “Figure 4.39 shows the frequency response of a dummy-head system equalized to a diffuse field, compared with the frequency response of a system equalized for the forward direction as in suggestion 1.”

Re: the so-called Blauert bands: https://science-of-sound.net/2016/01/about-jens-sound-localization-in-the-median-plane/

From Linkwitz at https://www.linkwitzlab.com/models.htm#H: "Around 3 kHz our hearing is less sensitive to diffuse fields. Recording microphones, though, are usually flat in frequency response even under diffuse field conditions. When such recordings are played back over loudspeakers, there is more energy in the 3 kHz region than we would have perceived if present at the recording venue and a degree of unnaturalness is introduced."

Also from Olive at https://www.psaudio.com/blogs/coppe...r-sean-olive-of-harman-international-part-one: "“Pinna gain is probably not the best term since the transfer function from a sound source to the eardrum is influenced by many anatomical parts besides the pinna. The well-known figure below summarizes the contributions in gain (dB) from the head, neck, torso, and external ear including the pinna. Together, they provide up to 17 dB gain at 2.7 kHz measured at the eardrum for a sound source located at 45 degrees to your left or right."

If you look at the so-called Shirley curve (page 152 of the second edition of Sound Reproduction, figure 9.7e), there does seem to be a dip below 2 kHz (maybe 1.8 kHz or so? My guess), a relative hump around 3-4 kHz or so (my guess), and another relative hump around 7 kHz or so (my guess), likely exacerbated in less reflective rooms. However, if you look at figure 9.7C, there is a substantial peak just above 2 kHz for a center speaker compared with the stereo speakers generating a phantom image. Consequently, the information above could explain why a phantom center seems more distant and elevated than an actual center to me.

 

[Thomas_A]

Okay, so, if testing in mono and stereo has shown speakers end up ranked in the same order, which from what is known they usually do, and if mono is more discriminating as in showing larger differences in listener preference, then how is listening in stereo a better test just because they get used in stereo? It would rarely change the ranking. If you use the results to find what speaker properties are most often desired, then what is gained by stereo testing? Unless the stereo test were more discerning or some property is desired for stereo that is different in kind or level in mono then why is stereo testing helpful? If speakers scoring highly in mono also do in stereo what is being missed by mono testing?

We all know that the center phantom image of a stereo setup is timbrally altered compared to a center mono source. Toole knows this, and therefore recommends adding a center channel or use multichannel. For stereo setups there are some choices that can be made (within limits of course) to at least reduce some of those stereo errors. What is lacking is proper trials and publications on the matter.

 

[youngho]

We all know that the center phantom image of a stereo setup is timbrally altered compared to a center mono source. Toole knows this, and therefore recommends adding a center channel or use multichannel. For stereo setups there are some choices that can be made (within limits of course) to at least reduce some of those stereo errors. What is lacking is proper trials and publications on the matter.

I think it's a little more complicated than that. For example, the perceived timbre (and measured frequency response) of a sound source also changes moving from 0 degrees to 30 degrees azimuth as in a standard stereo setup. See Shirley curve and Olive's comments I posted above, also https://www.madronadigital.com/perceptual-effects-of-room-reflecti (though the X-axis labelling in the head and pinnae transfer function seems a little inconsistent with https://headwizememorial.wordpress.com/2018/03/15/the-elements-of-musical-perception/).

 

[Blumlein 88]

I've done closely miked recordings in a dead space (not anechoic level dead). One recording for each musician. No reverb or any such. Placed speakers around similar to where the musicians might stand as a group. You get a startling step up in realism. It is a "they are here" realism, not a "you are there" realism. With no phantom imaging you don't want to mess with any response for being off axis because it sounds as it should. There will be unintended phantoms of a sort, but there is in reality too. Most of the time I discuss this people complain about how musician/instrument doesn't have the same directional characteristics as a speaker nor could I capture them correctly with a microphone. True, but it works wonderfully anyway.

With stereo important images are created in our minds where no physical sound is. I know this is not news to anyone. But unless it is done in the mix down, you cannot EQ or alter the side speakers to fix the center image without messing up the sound to the sides. Fixing one will alter the other. A center speaker mostly fixes this. One of the early bits of experiment at Bell labs was to put a row of microphones in front of an orchestra and have a speaker in a listening room at the same position as each microphone. Results were encouraging. They kept reducing the number of mikes and channels as obviously needing 10 or 20 channels was a no go then. The written description is 3 mikes and 3 speakers pretty well maintained those good results. Alan Blumlein's stereo with two channels took into account how our hearing works and is quite clever. It in time won out as 2 channels is easier than 3. So stereo it was, and stereo I suspect it will be for nearly all music consumption for a long time to come. Tests from the 1950's show 3 channels is much better for depth and accuracy of where things are heard vs 2 channels. Unfortunately, summing 2 channels into a center channel is a mixed bag according to those tests.

 

[MattHooper]

I've done closely miked recordings in a dead space (not anechoic level dead). One recording for each musician. No reverb or any such. Placed speakers around similar to where the musicians might stand as a group. You get a startling step up in realism. It is a "they are here" realism, not a "you are there" realism. With no phantom imaging you don't want to mess with any response for being off axis because it sounds as it should. There will be unintended phantoms of a sort, but there is in reality too. Most of the time I discuss this people complain about how musician/instrument doesn't have the same directional characteristics as a speaker nor could I capture them correctly with a microphone. True, but it works wonderfully anyway.

With stereo important images are created in our minds where no physical sound is. I know this is not news to anyone. But unless it is done in the mix down, you cannot EQ or alter the side speakers to fix the center image without messing up the sound to the sides. Fixing one will alter the other. A center speaker mostly fixes this. One of the early bits of experiment at Bell labs was to put a row of microphones in front of an orchestra and have a speaker in a listening room at the same position as each microphone. Results were encouraging. They kept reducing the number of mikes and channels as obviously needing 10 or 20 channels was a no go then. The written description is 3 mikes and 3 speakers pretty well maintained those good results. Alan Blumlein's stereo with two channels took into account how our hearing works and is quite clever. It in time won out as 2 channels is easier than 3. So stereo it was, and stereo I suspect it will be for nearly all music consumption for a long time to come. Tests from the 1950's show 3 channels is much better for depth and accuracy of where things are heard vs 2 channels. Unfortunately, summing 2 channels into a center channel is a mixed bag according to those tests.

All fair enough, and very interesting! I certainly find some of the obvious advantages with a center channel. I've posted it before and can't find the photo, but in recent audio shows there is a manufacturer (from china I think) who made a system devoted to reproducing orchestral music. The system uses a whole bunch of drivers in individual cabinets in a wide array, the idea being somehow to recreate more of the instrumental force/timbres or something. In any case, it got a fair number of good reviews in how well it reproduced orchestral pieces. (Wish I could remember the name).

But as mentioned before regarding center channels: the elephant in the room has always been practicality - designing and deploying a center channel that maintains the coherence with the other speakers, especially a center in how one is likely to be ever used in a domestic setting (hence typically horizontal designs). For all the drawbacks of mere stereo, the phantom central images can have a timbral and spatial coherence that beats a center channel.

Slightly adjacent to that topic: I've played some music videos in my home theater for some, one time with the full surround set up and center, another using just my stereo speakers, and they didn't even realize I'd changed to only stereo. The sonic imaging seemed solid and firmly mapped to the images.

 

[rynberg]

Remember that while many sources are recorded mono, once you have added stereo reverb (or similar effect/processing) to that source, you have created a stereo sound.
Likewise, plenty of electronic instruments are stereo.

Of course the intricacies all depend on how anything ends up being mixed, but the point is that it's too simple to think that a sound captured with a mono mic stays mono.
I add stereo effects and reverb to what were originally mono sources, thereby creating stereo files, all the time.

You left out the next part...and how are you creating a "stereo" sound? By panning the mono source between two channels...which is an adjustment to amplitude.

EDIT: you seem to be implying that I said that mono sounds stay mono, as in all entirely in one channel. I didn't actually say that (the opposite in fact), but of course that can and does happen. But panning it between two channels doesn't make it a true stereo source...

 

[Newman]

View attachment 341078
The sound quality rankings stayed the same.
The spatial quality ranking changed in going from mono to stereo (E moves into first place in spatial quality in stereo, though the margin may be small enough to be considered insignificant). Note that the "sound quality" scoring does not include the "spatial quality" scoring; those terms are applied to two separate sets of characteristics.

The degree of differentiation between the speaker rankings looks like it is consistently decreased in stereo, which would make stereo evaluation less valuable as a comparison technique.

Imo this three-speaker sample is far too small for those of us without access to a large body of raw data to draw reliable conclusions from. I believe @Floyd Toole when he says that the sound quality rankings never change when going from mono to stereo. It is not clear to me that he has said the same thing about the spatial quality rankings. Has he?

I was actually replying to a different Blumlein_88 post, where he specifically said he thought the ESL63 changed its preference ranking. So when I said it didn't, I was talking about the ESL63, which I think is BB on the chart.

Toole agreed with my interpretation of the stereo rankings, that you can't really say the rankings changed, because the error bars would be bigger than the differences in quality scores.

The stereo results in that graphic are basically ‘a wash’. In statistical terms, if error bars (confidence intervals) had been superimposed on the stereo chart, they would overlap. If experimenters conduct an experiment and the error bars overlap, it means they can’t validly say the experiment is showing anything at all. In other words, it’s a bad experiment, dominated by statistical noise/randomness. Yep, we are talking about using stereo for speaker evaluations.

cheers

 

[Newman]

Not sure what your point is here. I am not a native English speaker, so my apologies

Everyone who communicates outside their native language gets my respect, and no need to apologise.

...if my comments are sometimes inaccurate. Technically any speaker is a mono source 100% of the time, if we want to split hairs.

I agree with that, but like you suspect, that is not my point (which, BTW, is only a point that I picked up from Toole: it is not my own logic or argument). The point being, while playing in stereo, some of the music is coming out as a 2-speaker product, aka stereo, aka "double mono" (Toole's term), and some of the music is coming out panned quite hard to one side, or even entirely so. And that we are effectively listening to the latter music as a product of one speaker only, with the sonic attributes of a single speaker playing.

What I was trying to say I is that I am not explicitly designing speakers to be used as a single source of sound, so as a single mono speaker, and I suspect no customers use them that way either.

Yes, and I think Toole is telling us, as per the quotes I used, that when we play the speakers in stereo mode with music, a lot of program is hard-panned to one channel only, and we are hearing that program as a mono speaker, with the attributes of a single speaker.

Also, a lesson being that if we used a single speaker to evaluate speakers and chose the one that came out on top and utilised it for stereo music, we can't go wrong. But if we evaluate speakers as stereo, then two things can go wrong: (a) even quite large differences tend to be heavily masked and we might choose the one that actually has a lower stereo SQ but we couldn't pick it due to the 'experimental noise' in the stereo evaluation process, and (b) all the hard-panned program content is coming out with a significantly lower preference ranking, but we did our best to ignore and drown it during our stereo evaluation.

I am also not sure why you keep quoting studies and articles to me like you assume I have no idea what I am talking about.

I am simply providing evidence or legitimate authority to help explain my point. That is standard practice and not reserved for people who "have no idea". Please don't make it personal.

cheers

(to sigbergaudio): He's (Newman) quoting "studies" and "articles" (from an ancient publication) out of context, in order to attempt to make some dubious point/s that haven't been remotely articulated.

Seriously, just ignore the baiting would be my advice.

Thanks restorer-john, it's always good to know who here thinks that the best available science is so discredited that the studies, articles and books should be mocked as "studies" (in quote marks), "articles" and ancient publications (BTW my quotes were linked to the author’s postings from 2023). And good to know that you think that anyone referencing such findings should be thought of as baiting, and you recommend ignoring them.

Show us what you've got, then.

 

[Sancus]

From Toole: "Choisel and Wickelmaier (2007) found that listeners comparing a discrete center channel with a phantom center image generated
by a stereo pair in a normal room consistently rated the phantom image higher in perceptual dimensions of width, elevation, spaciousness, envelopment, and naturalness." I have some additional material regarding distance perception and a quote from JJ that I'll post later

That sentence doesn't show up in (my 3rd edition copy) of the book and those comparisons were made with a *single mono speaker*, so I'm not sure where you got the reference. And in any case, no one anywhere is making the case that a single mono center speaker is preferred over stereo. The minimum reasonable comparison would be 3-channel LCR vs LR stereo.

Toole's book very strongly states that a discrete center is better than phantom center. I don't think it's possible to make the case that it says the opposite without extreme cherry picking of quotes.

Are you talking about multichannel here, where you compare a real center channel as opposed to 2.x system where the center channel is mixed to the LR speakers?

Yes, I'm talking about multichannel with a real center channel compared to 2.x.

 

[Newman]

That sentence doesn't show up in (my 3rd edition copy) of the book and those comparisons were made with a *single mono speaker*, so I'm not sure where you got the reference.

pg 121 of the first edition

And in any case, no one anywhere is making the case that a single mono center speaker is preferred over stereo. The minimum reasonable comparison would be 3-channel LCR vs LR stereo.

You are exactly right, because if we extract the full statement by Toole in that quote, "Choisel and Wickelmaier (2007) found that listeners comparing a discrete center channel with a phantom center image generated by a stereo pair in a normal room consistently rated the phantom image higher in perceptual dimensions of width, elevation, spaciousness, envelopment, and naturalness. In a situation where the discrete center sound was unsupported by any sounds from other loudspeakers, this is consistent with expectations.
However, it is the task of the recording engineer to augment the spaciousness of a discrete center channel by using appropriately delayed and level-adjusted sounds sent to the left and right front channels and surround channels. If a phantom center is thought to have audible advantages, a real center channel, used in proper collaboration with processed signals delivered through other channels, has the potential to be better in every respect and much more flexible." (my emphasis)

 

[youngho]

That sentence doesn't show up in (my 3rd edition copy) of the book and those comparisons were made with a *single mono speaker*, so I'm not sure where you got the reference. And in any case, no one anywhere is making the case that a single mono center speaker is preferred over stereo. The minimum reasonable comparison would be 3-channel LCR vs LR stereo.

Toole's book very strongly states that a discrete center is better than phantom center. I don't think it's possible to make the case that it says the opposite without extreme cherry picking of quotes.

Yes, I'm talking about multichannel with a real center channel compared to 2.x.

The reference was from the second edition (also page 121) of Sound Reproduction, which I copied and pasted, which is why I included quotation marks. This comparison (there were others) was between a single mono speaker acting as a center channel and two speakers (stereo) playing multichannel content down-mixed to mono, hence a phantom center. You can find the paper here: https://web.archive.org/web/2007061...ckelmaier/pubs/ChoiselWickelmaier2007JASA.pdf

I was simply responding to this discussion:

I have found that to be true particularly of the center channel speaker. Ime a good two-channel system's phantom center image has more depth. Which doesn't really matter on movies, but imo that sense of depth adds to the experience with music videos. I have multiple customers who sold their center channel speakers because they no longer preferred to use them.

@Sancus : "Center channels sound substantially better than phantom center because of crosstalk cancellation. This is well-documented and studied. The "depth" you're talking about is exactly the problem, phantom center sounds indistinct and vague. That is an inaccuracy, not a beneficial spatial effect at all. Vocals are noticeably off from how they're supposed to sound and do sound in acoustic settings. They just come out inaccurate, and you can't fix it with EQ or by moving your speakers around. The best you can do is get as far away from the speakers as possible so you're mostly hearing reflected sound which reduces the interference, but that creates its own separate issues.

I've never seen any evidence that phantom center actually sounds better, just appeals to subjectivity."

There's probably no point in my commenting further.

 

[Blumlein 88]

For all the drawbacks of mere stereo, the phantom central images can have a timbral and spatial coherence that beats a center channel.

Don't think I agree with stereo having better spatial coherence or timbre. When that is the case it is because the center is a lesser or different speaker. Also it will usually be true over a more restricted area. It is not really true that stereo is better just that often 3 channel has compromised centers. Which is okay, having a fairly centered LP is something many of us live with all the time.