Now That Atmos Is Everywhere… Real vs. Phantom Center in a 5.1 Music-Focused Setup

[tengiz]

Atmos is turning up everywhere now – streaming, Blu-ray, even stereo playback - which got me wondering: in a music-focused 5.1 system, is a real center worth it, or is a phantom center just as good?

My setup started in the SACD/DVD-A days and I have a physical center that’s supposedly voice-matched to my mains (not really). I almost always prefer phantom for music - it blends better, keeps the stage seamless, and avoids tonal mismatch. A narrow sweet spot is fine for me, and I’m not chasing cinema SPL, so the "real center for headroom" argument doesn’t apply.

Atmos content can move audio anywhere, yet surrounds rarely match mains perfectly in real rooms. Dolby’s guidelines of course recommend tonal matching since the renderer doesn’t fix timbre differences, but Atmos still has to work on mismatched systems. So how much does matching really matter?

Has anyone compared? I haven’t heard Atmos music with truly flashy spatial effects yet. By “flashy” I mean something like the intro to Led Boots by Hiromi’s SonicBloom - plain stereo, yet spectacular, and for me it only works on speakers, not headphones.

 

[DVDdoug]

It's not just as good for 2 reasons.

- If you aren't sitting in the center you don't get as-good of a phantom center. (Your narrow sweet spot). Personally, I don't get a very precise phantom center illusion. It's sort-of vague. ...But it could be that I'm fooling myself, or maybe it's just my brain/ears, and I haven't done any controlled listening tests.

- There is something "strange" that happens with the stereo pair related to wavelength and the distance between your ears. There is a drop around 2kHz which of course affects dialog. With a stereo mix, of course the mixing engineer adapts to what he's hearing and the stereo is as it should be but there's no compensation for it when a real center channel is downmixed to stereo. (I believe I learned that from Toole's book.)

Another possible advantage is that a lot of people like to boost the center a few dB so they can better-hear the dialog, while lowering the overall level so they don't get blasted with effects.

 

[tengiz]

It's not just as good for 2 reasons.

- If you aren't sitting in the center you don't get as-good of a phantom center. (Your narrow sweet spot). Personally, I don't get a very precise phantom center illusion. It's sort-of vague. ...But it could be that I'm fooling myself, or maybe it's just my brain/ears, and I haven't done any controlled listening tests.

- There is something "strange" that happens with the stereo pair related to wavelength and the distance between your ears. There is a drop around 2kHz which of course affects dialog. With a stereo mix, of course the mixing engineer adapts to what he's hearing and the stereo is as it should be but there's no compensation for it when a real center channel is downmixed to stereo. (I believe I learned that from Toole's book.)

Another possible advantage is that a lot of people like to boost the center a few dB so they can better-hear the dialog, while lowering the overall level so they don't get blasted with effects.

I get that - but if an Atmos mix has objects panned between the center and one of the mains, any tonal mismatch in the physical center can be noticeable. I don’t hear it all the time when listening to the multichannel SACD layer - maybe I just don’t have many recordings that use that effect - but in controlled tests, it’s immediately obvious.

Atmos is “object-oriented” - at least that’s how I understand it - so it can essentially mix on the fly. That’s a big change from the old days, when humans hand-mixed the 5.1 and 2-channel layers. Back then, a mixer could deliberately avoid panning between the center and mains if they knew the center wasn’t a perfect tonal match. But with Atmos doing its own rendering, does it account for that? In most home setups, the center almost never matches the mains.

 

[slaweks]

I get that - but if an Atmos mix has objects panned between the center and one of the mains, any tonal mismatch in the physical center can be noticeable. I don’t hear it all the time when listening to the multichannel SACD layer - maybe I just don’t have many recordings that use that effect - but in controlled tests, it’s immediately obvious.

Atmos is “object-oriented” - at least that’s how I understand it - so it can essentially mix on the fly. That’s a big change from the old days, when humans hand-mixed the 5.1 and 2-channel layers. Back then, a mixer could deliberately avoid panning between the center and mains if they knew the center wasn’t a perfect tonal match. But with Atmos doing its own rendering, does it account for that? In most home setups, the center almost never matches the mains.

I am not sure how it works, I am using Marantz Av 10, on some recordings the center speaker is seemingly not used at all, while on other plays prominent role. Also front wides, on most recordings play just background reflections, on some are completely mute, and in minority cases, play nicely separate instruments, so you have the whole wall of sound coming from 5 speakers.

 

[BJL]

Atmos content can move audio anywhere, yet surrounds rarely match mains perfectly in real rooms. Dolby’s guidelines of course recommend tonal matching since the renderer doesn’t fix timbre differences, but Atmos still has to work on mismatched systems. So how much does matching really matter?

I have a perfectly matched set of seven speakers for the bed channels, and yes it makes a big difference, not just with Atmos, but also with 5.1/7.1 music discs (and movies/tv for that matter).

If you do not want to replace your center channel speaker, the other approach is to run Dirac or similar, which should smooth out the differences. Even though my speakers are matched, I still use Dirac to address room interactions and for integration of the bed speakers with the Atmos speakers.. Dirac elevates the audio quality of my system from excellent to sublime.

My Atmos speakers (heights and front wide) are perfectly matched as a group, but do not match the bed speakers. Dirac takes care of that nicely.

And going back to your initial question, I have found the center channel to be substantively significant in many music recordings (and of course for movies), and when I once tried a phantom center, I found it to be sorely lacking.

I believe that the solution to your issue is to purchase a center speaker that is identical to your from left and right, and of course, the optimal system uses identical speakers in all positions where that is possible, or alternatively try room correction/speaker eq software to smooth out the differences.

 

[tengiz]

I am not sure how it works, I am using Marantz Av 10, on some recordings the center speaker is seemingly not used at all, while on other plays prominent role. Also front wides, on most recordings play just background reflections, on some are completely mute, and in minority cases, play nicely separate instruments, so you have the whole wall of sound coming from 5 speakers.

If it stays separate or static in height/surround/wides/etc speakers that don’t match the mains, I can understand it working. But pairing arbitrary speakers for panning — that doesn’t seem workable in any realistic scenario.

 

[tengiz]

I have a perfectly matched set of seven speakers for the bed channels, and yes it makes a big difference, not just with Atmos, but also with 5.1/7.1 music discs (and movies/tv for that matter).

If you do not want to replace your center channel speaker, the other approach is to run Dirac or similar, which should smooth out the differences. Even though my speakers are matched, I still use Dirac to address room interactions and for integration of the bed speakers with the Atmos speakers.. Dirac elevates the audio quality of my system from excellent to sublime.

My Atmos speakers (heights and front wide) are perfectly matched as a group, but do not match the bed speakers. Dirac takes care of that nicely.

And going back to your initial question, I have found the center channel to be substantively significant in many music recordings (and of course for movies), and when I once tried a phantom center, I found it to be sorely lacking.

I believe that the solution to your issue is to purchase a center speaker that is identical to your from left and right, and of course, the optimal system uses identical speakers in all positions where that is possible, or alternatively try room correction/speaker eq software to smooth out the differences.

I haven’t come across any room correction that delivers truly seamless transitions in panning tests - except between the mains. Correction gets you part of the way there, but as soon as sounds move across channels, the mismatches show up.

That makes me wonder: does Dirac actually handle this better than Audyssey? Maybe its mixed-phase approach - or whatever they call it - gives it an edge, but how much does that really matter if the center itself isn’t a tonal match? That seems like the bigger problem.

The rest of the speakers outside of mains + center always seem inherently disadvantaged by placement. If your experience with Dirac is that it still feels satisfactory, then maybe I should give it a try - my receiver is Dirac-ready. But their offer is multi-tiered: there’s the basic version, then the upgrade with better bass management, and then another tier beyond that. What do you use in your setup?

As for me, I don’t have space for a center identical to the mains - they’re simply too tall. So I’m stuck with the same choice every time: a “dedicated” center that still doesn’t blend (despite the manufacturer’s claims that it’s the matching model), or no center at all.

 

[BJL]

I haven’t come across any room correction that delivers truly seamless transitions in panning tests - except between the mains. Correction gets you part of the way there, but as soon as sounds move across channels, the mismatches show up.

That makes me wonder: does Dirac actually handle this better than Audyssey? Maybe its mixed-phase approach - or whatever they call it - gives it an edge, but how much does that really matter if the center itself isn’t a tonal match? That seems like the bigger problem.

The rest of the speakers outside of mains + center always seem inherently disadvantaged by placement. If your experience with Dirac is that it still feels satisfactory, then maybe I should give it a try - my receiver is Dirac-ready. But their offer is multi-tiered: there’s the basic version, then the upgrade with better bass management, and then another tier beyond that. What do you use in your setup?

As for me, I don’t have space for a center identical to the mains - they’re simply too tall. So I’m stuck with the same choice every time: a “dedicated” center that still doesn’t blend (despite the manufacturer’s claims that it’s the matching model), or no center at all.

I have the standard verion of Dirac, i.e. full range but no dynamic bass management or other bells and whistles. It treats the sub as another speaker, level matching, EQ, timing, phase, but no bass management features. Dirac does recognize it as a subwoofer. Maybe you can get a demo?

For the center, I have my TV on a stand that raises it so the bottom of the TV is 42 inches above the floor, my center sits right under the TV with the top of the speaker just barely below the TV. In the past I tried a couple of so-called center channel speakers and returned them because, as you point out, they generally speaking do not match the other speakers. Perhaps very high end centers do, I wouldn't know about that.

I have never used Audyssey. My experience with Dirac in two systems with completely different speakers, is that the Dirac software controls the tonal quality, so that my newer (and superior) speakers sound similar, tonally, to the previous, less good, set of speakers. Of course the newer set are more detailed and other improvements. But the tonal quality is similar. It's more that the newer speakers require much less correction than the older speakers. See if you can get a demo, or return privilege. You might be pleasantly surprised.

 

[RDoc]

It seems to me that to test tonal quality, in addition to identical inputs and any needed volume matching, the two speakers need to be very close together or the room acoustics will dominate. Has anyone tried this?

 

[slaweks]

It seems to me that to test tonal quality, in addition to identical inputs and any needed volume matching, the two speakers need to be very close together or the room acoustics will dominate. Has anyone tried this?

I would hope something like Dirac Live would take care of volume matching.

 

[tengiz]

It seems to me that to test tonal quality, in addition to identical inputs and any needed volume matching, the two speakers need to be very close together or the room acoustics will dominate. Has anyone tried this?

But if there’s a tonal mismatch - even after the correction is applied - volume matching becomes ambiguous, right? Loudness perception depends not only on overall SPL but also on spectral balance, so two sources that measure the same can still sound different because of their frequency responses.

And since most automatic room-correction systems rely on spatial averaging, that only adds more uncertainty. Which is why I’m not convinced tonal matching between different speakers can ever be more than a partial fix - helpful to a degree, but ultimately an elusive goal.

 

[tengiz]

I would hope something like Dirac Live would take care of volume matching.

They do. But like almost everything related to hearing, it has to rely on aggregate metrics. Take SPL meter weighting, for example: you have a continuum of frequencies and magnitudes in the input, and it all has to be collapsed into a single value. The problem is that there are countless ways to distribute energy across the spectrum that yield the same SPL number - yet they do not sound the same.

They might also use something more elaborate than a single-number comparison - for example, analyzing and comparing power spectrum shapes in the most critical frequency ranges instead of just overall levels. But even then, they cannot get around the spatial dimension. Multiple measurement points still have to be reduced to one decision metric - whether scalar or vector - which means that in the end, the data must be collapsed into something simpler.

 

[Soundmixer]

I get that - but if an Atmos mix has objects panned between the center and one of the mains, any tonal mismatch in the physical center can be noticeable. I don’t hear it all the time when listening to the multichannel SACD layer - maybe I just don’t have many recordings that use that effect - but in controlled tests, it’s immediately obvious.

Atmos is “object-oriented” - at least that’s how I understand it - so it can essentially mix on the fly. That’s a big change from the old days, when humans hand-mixed the 5.1 and 2-channel layers. Back then, a mixer could deliberately avoid panning between the center and mains if they knew the center wasn’t a perfect tonal match. But with Atmos doing its own rendering, does it account for that? In most home setups, the center almost never matches the mains.

There is quite a bit of misunderstanding here. Atmos does not "mix on the fly". It is a scalable system that adapts to all of the Atmos-certified speaker setups.
Mixers don't deliberately avoid panning between the center and mains, as they don't know what your center or mains are, how capable they are, or if you are missing a center. Mixers don't know if your center is a tonal match or not; how can they?

It is a mistake to think Atmos remixes itself on the fly to adapt to your setup. When you tell the processor you have no center, it will shunt all center channel information evenly to the L/R mains. The problem with this is that there is an HRT cancellation that occurs between 2-4Khz, and that will affect dialog intelligibility when there is a lot of action in the other channels. Not only that, if you are not sitting dead center, the dialog will pull to whatever channel you are sitting closest to. This destroys the immersive effect of Atmos.

 

[dlaloum]

Atmos is turning up everywhere now – streaming, Blu-ray, even stereo playback - which got me wondering: in a music-focused 5.1 system, is a real center worth it, or is a phantom center just as good?

My setup started in the SACD/DVD-A days and I have a physical center that’s supposedly voice-matched to my mains (not really). I almost always prefer phantom for music - it blends better, keeps the stage seamless, and avoids tonal mismatch. A narrow sweet spot is fine for me, and I’m not chasing cinema SPL, so the "real center for headroom" argument doesn’t apply.

Atmos content can move audio anywhere, yet surrounds rarely match mains perfectly in real rooms. Dolby’s guidelines of course recommend tonal matching since the renderer doesn’t fix timbre differences, but Atmos still has to work on mismatched systems. So how much does matching really matter?

Has anyone compared? I haven’t heard Atmos music with truly flashy spatial effects yet. By “flashy” I mean something like the intro to Led Boots by Hiromi’s SonicBloom - plain stereo, yet spectacular, and for me it only works on speakers, not headphones.

A comment on Tonal matching ....

Basically tonal / Timbre matching is related to the frequency response profile or "voicing" of the speakers.

If your system uses RoomEQ tools such as Dirac Live, you can adjust the F/R (Target curve) for individual speakers or groups of speakers (channels / pairs)

Hence you can apply the Timber/Tone/Voicing you want.

How I use this:

I like the "sound" of my mains (and matching center) - so I have used their measured room curve as my template, created a matching target curve, and applied it to my surround / height speakers.... now they are "timbrally" matched.

Fly in the ointment - every speaker has its quite individual resonances... you cannot correct for resonances, and if they are substantive then they can become noticeable and affect the timbre of the speakers - it becomes noticeable when a sound moves around the room from one channel to another... and the sound varies as it moves.... even though they are as matched as possible the resonances/flaws can give them away.

Note: this applies to differing models from the same manufacturer, purportedly "matched", just as much as it applies to mismatched speakers from different manufacturers.... hence the ideal is to have identical speakers all around - often simply not possible for most of us!

I'm very happy with my results using the fronts as my template and matching the others to the fronts - all the speakers are pretty decent, with minimal apparent resonances, so the outcome is very good (Front Gallo Nucleus Reference 3.2, center Gallo Nucleus Reference AV, surrounds B&O Penta, Heights Gallo Nucleus Micro). - My setup is 5.1.4.

Most of my listening / viewing is 5.1 or stereo material - Atmos features more rarely.

 

[tengiz]

Mixers don't deliberately avoid panning between the center and mains, as they don't know what your center or mains are, how capable they are, or if you are missing a center. Mixers don't know if your center is a tonal match or not; how can they?

Yes, that’s exactly one of my questions - if mixers aren’t carefully avoiding "unsafe" pans when played back across the thousand-and-one unknown yet "certified" almost always subpar "non-standard" home layouts, then how does this whole scheme even work in practice?

It is a mistake to think Atmos remixes itself on the fly to adapt to your setup. When you tell the processor you have no center, it will shunt all center channel information evenly to the L/R mains.

A-ha - so if the fallback is just dumping a missing channel into L/R, even when other speakers would make more sense (like routing Back Surrounds into Surrounds), then the system’s simpler - and maybe safer - than it first looks. But that raises the real question: what’s actually “object-oriented” about this? From your description it still just shuffles streams to whatever speakers exist - basically a downmix, only dressed up with “objects” instead of channels. Is that really all there is?

And if scalability just means breaking things into object streams, then what’s left of the big pitch about spatial metadata and motion? Because if objects really carry that, the renderer still has to grind through levels, delays, and maybe even phases in real time to map it to a given setup. In other words, isn’t this just the same old math in a fancier box with better branding?

The problem with this is that there is an HRT cancellation that occurs between 2-4Khz, and that will affect dialog intelligibility when there is a lot of action in the other channels. Not only that, if you are not sitting dead center, the dialog will pull to whatever channel you are sitting closest to. This destroys the immersive effect of Atmos.

And so does a typical subpar home setup. Unless Atmos handles this in a more robust way than it seems on the surface - why bother at all?

 

[sfogg]

Atmos is turning up everywhere now – streaming, Blu-ray, even stereo playback - which got me wondering: in a music-focused 5.1 system, is a real center worth it, or is a phantom center just as good?

My setup started in the SACD/DVD-A days and I have a physical center that’s supposedly voice-matched to my mains (not really). I almost always prefer phantom for music - it blends better, keeps the stage seamless, and avoids tonal mismatch. A narrow sweet spot is fine for me, and I’m not chasing cinema SPL, so the "real center for headroom" argument doesn’t apply.

I have always been the opposite in that a real center is better than a phantom center. Beyond the narrow sweet spot a real center avoids the comb filtering on vocals that occur with a phantom center channel between L/R. That doesn't occur in the real world with a vocalist but it is so prevalent with 2 channel stereo removing the comb filtering takes a bit of time to get accustomed to the difference. Well matching center is of course very important (my L/C/Rs are identical) as is time alignment. If you are using surround processing then good front steering is also really important to bring the center into play without collapsing everything into the center speaker.

 

[tengiz]

If your system uses RoomEQ tools such as Dirac Live, you can adjust the F/R (Target curve) for individual speakers or groups of speakers (channels / pairs)

If we’re talking about full-band room correction (the autocorrect, unironically, suggests “room corruption” ) - like Audyssey MultEQ-XT, which by default tries to eliminate FR differences between speakers - shouldn’t that be enough? And yet, in almost 20 years I’ve never once managed to get a truly seamless blend between different models of mains and their supposedly “voice-matched” centers.

So how exactly would manual adjustments help here?

 

[tengiz]

I have always been the opposite in that a real center is better than a phantom center. Beyond the narrow sweet spot a real center avoids the comb filtering on vocals that occur with a phantom center channel between L/R. That doesn't occur in the real world with a vocalist but it is so prevalent with 2 channel stereo removing the comb filtering takes a bit of time to get accustomed to the difference. Well matching center is of course very important (my L/C/Rs are identical) as is time alignment. If you are using surround processing then good front steering is also really important to bring the center into play without collapsing everything into the center speaker.

I get these arguments (though stereo still somehow delivers its magic), but I can’t square them with the practical reality: a perfectly matched center is a luxury, not the norm. And yet Atmos is marketed not as a premium niche, but as something universal - which makes the gap between promise and practice hard to ignore.

 

[dlaloum]

If we’re talking about full-band room correction (the autocorrect, unironically, suggests “room corruption” ) - like Audyssey MultEQ-XT, which by default tries to eliminate FR differences between speakers - shouldn’t that be enough? And yet, in almost 20 years I’ve never once managed to get a truly seamless blend between different models of mains and their supposedly “voice-matched” centers.

So how exactly would manual adjustments help here?

Well there are a couple of things....

If the speakers aren't identical all around (not just from the same manufacturer with a claim to timbral matching!) - then there will be differences.

As we live in an imperfect world (well most of us) - we accept the imperfections and do our best to minimise them!

You can take a generic target curve, and apply it equally to all speakers... this will reduce timbral variation, but as I mentioned earlier, even in the best case, resonance variations between differing speakers will make them potentially identifiable.... it's that imperfect world thing!

My alternative is to minimise processing on the matched L/C/R - which are a set of speakers I chose because I liked their voicing, and it works well in my space.

So I chose to set my target curve, using their measured response curve - that means that Dirac does very little on the L/C/R - and more on the surrounds and heights.... not a negative thing. - Dirac still does its impulse response/phase thing - and it results in a distinct audible improvement - but it minimises the messing with the response curve.

Other than specifying a "Non-standard" target curve, I don't make manual adjustments - I let Dirac do its thing.

I believe something similar can be achieved using Audyssey and the related add-on apps (which are needed for tweaking the target curve) - how easy or hard it is in the Audyssey realm, I don't know as I no longer have an Audyssey AVR, and when I did have one, the tweaking was not an option.

P.S. I find that my Gallo Nucleus Reference 3.2 L/R and Gallo Nucleus Reference AV center, are very well matched timbrally - the F/R measurements using both Dirac and REW agree....they pretty much have the exact same frequency response curve above roughly 120Hz - which helps.

My setup is 5.1.4 - and my surrounds are B&O Penta speakers (the early passive version) - which coincidentally has a quite similar F/R to the Gallo's... so there isn't too much adjustment required on Dirac's part to match them to the target curve! - Heights are Gallo Nucleus Micro's.... and that's where the most adjustment happens, not surprisingly.

 

[sfogg]

"Has anyone compared? I haven’t heard Atmos music with truly flashy spatial effects yet. By “flashy” I mean something like the intro to Led Boots by Hiromi’s SonicBloom - plain stereo, yet spectacular, and for me it only works on speakers, not headphones."

I just jumped back to your original question above and did it by comparing your link in stereo vs that in surround. My 11 surround speakers are all identical to each other but are not the same speaker as my L/C/R or stage height LCRs.

Sitting in the sweet spot, in stereo, the phase tricks are fun with the sound jumping outside of L/R and moving from your side to side with a little bit of it sounding like it is just over your shoulder as it is passing through your head. Those phase tricks can work well to the sides but they don't image as well to the rear, that is why it sounds like it is sort of going through your head or just behind you.

I took that 2 channel source and ran surround processing on it to expand it out to 18 channels. Playing that back in surround has those sounds smoothly circling the room, without the kind of over your shoulder/through you head feeling. No need for the head in a vice to get it either. It is much more enveloping, open and immersive. And that is with surrounds that are not the same as the L/C/R.