The “Real” Speaker Preference Test

[Floyd Toole]

tl;dr answer for thread: No

I suspect that the answer is more like "yes", but not because the public recognizes more neutral sound quality in typical listening conditions when making their choices. Such opinions, as is well proven, are influenced by many factors having nothing to do with the sound arriving at the ears. This is the decades old process by which manufacturers sold everything from junk to jewels to an uninformed public. There was no option.

The growing reality is that increasing numbers of loudspeakers of all kinds are being designed to follow the guidance of the research as embodied in the CTA2034 standard. This yields more accurate, timbrally neutral loudspeakers. This can be seen in the spinorama measurements done by Amir and Erin, and which competent manufacturers now are beginning to show in their specifications. This applies to smart speakers, TV sound bars, consumer audio and professional monitor loudspeakers - at all price levels. Interestingly, a side effect is that there is a greater chance of hearing the recorded art as it was created.

So, whether they know it or not, increasing numbers of people are buying better loudspeakers because of the NRCC and Harman research. BTW, the core research was done at the National Research Council of Canada.

 

[Floyd Toole]

Do you think I'm being unreasonable? I'm sure I have my blind spots.

Don't we all! Before commenting further I will read his thesis again - it has been many years since I did.

Sounds arriving from different directions have different timbres because HRTFs are different for each incident angle. In that sense there is definitely some real time compensation going on because we subconsciously attribute those timbral variations with the perceived direction and simply say: " that's what it sounds like from that direction" - no problem. That is a personal adaptation process because we all have distinctive HRTFs. Those differences can be huge.

The timbral differences caused by acoustical crosstalk in stereo are subtle. Because of head shadowing really only the 2 kHz dip is likely to be a significant issue. When auditioned with pink noise as one moves very slightly left and right of the sweet spot it is obvious. With program it is less audible.

Anyway more to come later. Cheers,

 

[Duke]

Don't we all! Before commenting further I will read his thesis again - it has been many years since I did.

Thank you for replying, but PLEASE don't feel obligated to dig up and read Theile's paper!

(Here's a link though, just in case: https://hauptmikrofon.de/theile/1980-2_Diss._Theile_englisch.pdf)

 

[MattHooper]

Rumsey reports that the least coloration comes from using only two channels! (Go to 47:16 in the video embedded in my previous post.) So, maybe two-channel sound quality doesn't necessarily inherently suck compared to using a dedicated center-channel speaker.

Regarding the perceived spatial quality superiority of the phantom center configuration, that too might be possible; quoting from the King et al paper you cited above: “Research shows that phantom center images are often preferred over a discrete center, because of the added spaciousness, envelopment, etc.” So, maybe two-channel has desirable spatial-quality attributes compared to using a dedicated center channel speaker.

I’ve mentioned this before, but…

Since my two channel speakers share the room with my surround system I’ve done comparisons switching between live concert music with good spatial qualities (both electronic and acoustic), both just listening to music, and also watching with picture on the projection screen.

I have a very good 7.0 surround system to compare with my two channel speakers.

My subjective findings were that the surround system won on ultimate surround immersion, but the two channel speakers won on other qualities - coherency of timber and spatial qualities, detail, greater spatial and imaging realism. When watching concerts I loved the immersiveness of the surround, but the two channel speakers pulled well out into the room gave a more realistic sense dissolving into the acoustics of the recording. It was that much more real sounding to me.

This of course, was not a perfectly controlled apples to apples comparison, and for me there are sonic advantages to being able to position the two channel speakers out into the room and closer to me:

(You don’t normally get that type of scenario with the placement of surround speakers. A KEF Blade based LCR system like Kal Rubinson’s would probably be a pretty rare type of exception)

 

[Duke]

My subjective findings were that the surround system won on ultimate surround immersion, but the two channel speakers won on other qualities - coherency of timber and spatial qualities, detail, greater spatial and imaging realism. When watching concerts I loved the immersiveness of the surround, but the two channel speakers pulled well out into the room gave a more realistic sense dissolving into the acoustics of the recording. It was that much more real sounding to me.

This is pretty much what was observed in the informal test I mentioned. The set-up was a bit different, three identical speakers positioned well out into a fairly large room, and a trinaural processor with a bypass switch that allowed instant switching. The processor used was linear, in that it does not change the net volume level when switching between trinaural and bypass modes. So there was a fair amount of apples-to-apples going on.

In stereo mode the spatial presentation was realistic, with a deep soundstage. In trinaural mode the depth of the soundstage right in the middle was constrained to approximately the distance to the center-channel speaker, maybe a little deeper, but the net result seemed unrealistic to the point of being distracting in comparison with the soundstage in stereo mode. Surround speakers were not used. If there was a coloration difference none of us caught it, and I had gone into the test expecting to hear an audible difference arising from the presence or absence of the 2 kHz comb filter dip. (Francis Rumsey's lecture is the first time I have come across anything validating our not perceiving the presumptive timbral superiority of using a dedicated and identical center-channel speaker.)

This one minor and EQ-fixable timbral difference was observed: When listening to music videos through the drop-down projection screen in trinaural mode, there was a slight reduction in high frequency energy presumably because the screen was attenuating a little bit in the top octave. The screen did not cover the left and right speakers, and the top-end energy seemed to be fully present in phantom center mode with the screen down.

This little experiment was an unexpected point of inflection for me, as I had been working on a design that could be used for a center-channel speaker as well as left & right main speakers. It was an 8" MCM (midwoof-coaxial-midwoof) using prosound drivers, and its fairly wide and well-behaved radiation pattern was the same in both vertical and horizontal orientations down to the crossover to the two midwoofers, ballpark 500 Hz. That project got shelved.

 

[Jukebox]

This is pretty much what was observed in the informal test I mentioned. The set-up was a bit different, three identical speakers positioned well out into a fairly large room, and a trinaural processor with a bypass switch that allowed instant switching. The processor used was linear, in that it does not change the net volume level when switching between trinaural and bypass modes. So there was a fair amount of apples-to-apples going on.

In stereo mode the spatial presentation was realistic, with a deep soundstage. In trinaural mode the depth of the soundstage right in the middle was constrained to approximately the distance to the center-channel speaker, maybe a little deeper, but the net result seemed unrealistic to the point of being distracting in comparison with the soundstage in stereo mode. Surround speakers were not used. If there was a coloration difference none of us caught it, and I had gone into the test expecting to hear an audible difference arising from the presence or absence of the 2 kHz comb filter dip. (Francis Rumsey's lecture is the first time I have come across anything validating our not perceiving the presumptive timbral superiority of using a dedicated and identical center-channel speaker.)

This one minor and EQ-fixable timbral difference was observed: When listening to music videos through the drop-down projection screen in trinaural mode, there was a slight reduction in high frequency energy presumably because the screen was attenuating a little bit in the top octave. The screen did not cover the left and right speakers, and the top-end energy seemed to be fully present in phantom center mode with the screen down.

This little experiment was an unexpected point of inflection for me, as I had been working on a design that could be used for a center-channel speaker as well as left & right main speakers. It was an 8" MCM (midwoof-coaxial-midwoof) using prosound drivers, and its fairly wide and well-behaved radiation pattern was the same in both vertical and horizontal orientations down to the crossover to the two midwoofers, ballpark 500 Hz. That project got shelved.

Thanks for sharing such a detailed account of that experiment—it resonates with my findings also.

I’ve been testing dedicated center vs. phantom center in my own system for about 5 years now, and despite what acoustic theory and textbooks preach about the absolute superiority of a dedicated center channel, my real-world results have been surprisingly split:

• The Center Channel Advantage: Yes, a physical center anchors everything flawlessly, especially for off-axis listening. Dialogue and center-focused elements stay rock-solid no matter where you sit. However, depth and airiness take a noticeable hit with the center engaged
• The Phantom Center Advantage: When I pull my L+R speakers well out into the room and run them stereo (phantom center), the soundstage opens up dramatically with a depth, three-dimensionality, and sense of space that a center speaker just seems to collapse.

For a long time, I actually ran a 4.2.4 setup (no center) precisely because I couldn't part with that spaciousness. Recently, I decided to give the center another shot and pushed my L+R mains back inline with the center speaker, but even then, I’m still not convinced it’s better.

What caught my attention in your comment is that you tested all three identical speakers pulled well out into the room and still found the phantom center superior in terms of realistic depth! I actually haven't tried bringing all three speakers out into the room together yet , so that’s a test I definitely need to run next.

It’s fascinating (and reassuring) to see that even with three identical, well-positioned speakers and a proper linear processor, human perception of soundstage depth often favors the phantom image over the physical anchor.
Appreciate you sharing this—it gives me some great food for thought for my next round of tweaking!

 

[Cote Dazur]

The ultimate Preference Test is whether people....

The ultimate preference test is if I prefer something or not, what other people do, for what ever reason, in what ever circumstances, may or may not align with my preference.
The speaker performance, contrary to any other link in sound chain, is greatly affected by many factors, you always end up with a compromise.
Starting with a well measuring speaker seems like the smart thing to do, but does in no way warranty you will prefer it if installed improperly in a bad sounding room.

 

[kyuu]

Starting with a well measuring speaker seems like the smart thing to do, but does in no way warranty you will prefer it if installed improperly in a bad sounding room.

There's no reason to think one would prefer a speaker that measures poorly in that situation. At least if the speaker measures well you can be fairly certain the speaker is not the issue and you need to address the setup and room. If the speaker measures poorly, the pooch is screwed from the start.

 

[Cote Dazur]

There's no reason to think one would prefer a speaker that measures poorly in that situation. At least if the speaker measures well you can be fairly certain the speaker is not the issue and you need to address the setup and room. If the speaker measures poorly, the pooch is screwed from the start.

it seem to me like we are saying the same thing.

 

[kyuu]

it seem to me like we are saying the same thing.

Maybe, perhaps I wasn't understanding you correctly.

 

[Blueprint]

I wouldn’t expect a center channel to outperform a phantom center if the source is stereo and the listener is central, since the HRTF-related phantom issue is already accounted for in the stereo mixing and stereo’s inherent symmetry would likely sound more cohesive.

 

[NTK]

...
I’ve been testing dedicated center vs. phantom center in my own system for about 5 years now, and despite what acoustic theory and textbooks preach about the absolute superiority of a dedicated center channel, my real-world results have been surprisingly split:
...

How did you come up with the signals for the center channel? You'd have, at the minimum, use an upmixer, if not an original multichannel mix.

If you just sum the left and right channels to create the center, and leave the L&R alone, or maybe with some channel level adjustments to maintain the same output energy between 2 and 3 channels, you'd end up with a narrower sound stage. The summed center will pull any sound that's off center towards the center.

To take advantage of the center channel (from stereo contents), a good upmixer is a must. The best technique is to limit any sound to no more than 2 channels, e.g. when a sound is panned to somewhere between the center and right, it is only reproduced by the center and right speakers, and the left speaker doesn't contribute (or contributes very very little). The process for this "directional decoding" to extract the center channel from a stereo recording is far from trivial.

 

[Jukebox]

How did you come up with the signals for the center channel? You'd have, at the minimum, use a upmixer, if not an original multichannel mix.

Sorry for not mentioning
I am talking here only about stereo music (since it represents for 90% of the content used) up mixed with Auro or Dolby via Denon AV10H

 

[Duke]

• The Center Channel Advantage: Yes, a physical center anchors everything flawlessly, especially for off-axis listening. Dialogue and center-focused elements stay rock-solid no matter where you sit. However, depth and airiness take a noticeable hit with the center engaged
• The Phantom Center Advantage: When I pull my L+R speakers well out into the room and run them stereo (phantom center), the soundstage opens up dramatically with a depth, three-dimensionality, and sense of space that a center speaker just seems to collapse.

... What caught my attention in your comment is that you tested all three identical speakers pulled well out into the room and still found the phantom center superior in terms of realistic depth! I actually haven't tried bringing all three speakers out into the room together yet , so that’s a test I definitely need to run next.

Thanks for posting about your experience!

The test I described used fairly small bookshelf speakers, not my original design, though I did modify them a bit.

I normally design for time-intensity trading which imo significantly improves the off-centerline imaging in phantom center mode. No, the center image is not as anchored to the screen as with a dedicated center channel, but with eyes closed from most listening/viewing locations the dialogue still comes from the screen.

Imo the retained benefit is as you describe: "the soundstage opens up dramatically with a depth, three-dimensionality, and sense of space that a center speaker just seems to collapse."

 

[dped90]

As for the suppression of coloration when listeners are aware of the loudpeakers, surely this is easily disproven by substituting an identical centre loudspeaker for a phantom image created by a pair of identical loudspeakers. This is the situation faced by recording engineers creating programs for multiple playback formats. Recording engineers who sit in the sweet spot are challenged by the timbral mismatch between a real center loudspeaker and a phantom centre image.

I was discussing similar concerns with today with Duke. Wondering is how problematic would it be if I added an identical center two-way speaker (https://at-horns.eu/A460D.html https://josephcrowe.com/blogs/news/altec-416-8b-in-100l-sealed ) but the side speakers were something like this? https://philharmonicaudio.com/products/ht-tower

And where mono and stereo source material might be 5.1 upmixed with software like these?
www.perfectsurround.com/ or https://nugenaudio.com/haloupmix/

I started this thread to ask some in the recording business for their feedback. Some seemed were rather put off but I did receive some useful feedback. https://gearspace.com/board/vr-virt...pmixing-software-5-1-mono-stereo-sources.html

You might want to click on those youtubes, especially if you have access to a multichannel system. There's definitely some serious noise and/or clipping distortion in the guitar bridge but I'd like to know how that actual mono to 5.1 mix went.

I wouldn't be pursuing any kind of 5.1 system anytime soon but Duke referred me your dialogs with him here on phantom channel coloration and hope you would share your thoughts on what you might know about such software and especially about the apparent advantages of an identical three speaker/channel system.

Thanks for this great sub topic.