The Problem with Many Center Channel Speakers (by Erin)

[bjmsam]

The benefits of a centre channel are dealt with in great detail in @Floyd Toole 's book Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms

Yes, but consider that in the context of his position on early reflections (per @Ethan Winer).

I can understand the potential benefits of adding a centre channel from an imaging/soudstanging perspective but that simulation does not bode well for the tonal accuracy side of things...

And why stop at one? If I had a center channel speaker for my 120" screen, would phantom "center" not still be employed for sounds between L and C and between C and R?

 

[MattHooper]

That was very interesting!

Center channels have been getting grief for a long time, so while the gist wasn't new to me, the specific info, graphs and tests certainly added more detail. Excellent video.

Interesting to see the 3 way WTMW design come out on top. It matches my anecdotal experience.

When I was setting up my home theater many years ago I'd decided on using Hales Transcendence T1 stand mounted speakers for my L/R, as Ioved their clean, clear, spacious and timbrally convincing sound. They were to flank my projection screen and I needed a center channel. Unfortunately Hales had become defunct at that point, not long after they had introduced speakers designed for home theater, and very few of their center channels had made it to the market. I tried various center channels, including coaxial speakers, even various stand mounted designs (even Waveform Mach MC egg-shaped speakers which have excellent off-axis performance). Nothing really produced a seamless match.

Finally I managed to find a rare-as-hen's-teeth Hales Transcendence Center channel speaker, designed to specifically match the rest of the Transcendence line speakers. It's a big ol' sealed box 3 way WTMW, seen in the photo below (a photo I grabbed from the web, not from my own set up):

It was just the ticket! The timbral and spatial match was virtually seamless with the L/R T1 speakers! Everything from voices to any other sounds that would pan through the scene would remain beautifully consistent in character. And I've remained impressed by how well the dialogue seems to map to the image even though the center channel is just below the screen (angled up to the listener at the sofa). And I'm super sensitive to a mismatch of dialogue to the screen. It's one of the things that immediately turn me off in many home theater systems (especially of course home theater in a box systems...but often enough in "real" home theater systems).

So I suppose Erin's video helps explain to me somewhat why this L/C/R system has worked so well for me.

BTW, that also brings up one of my personal pet peeves in home theater set ups, especially the version of "home theater" that includes any flat panel system with surround sound in someone's living room or whatever. That is: a mismatch between the scale of the sound with the scale of the image. When home theater sound was becoming really big it was so common in box store demos to see a 42" - 50" flat panel paired with large Paradigm or Klipsch speakers, and even big floor standing surround speakers. I've seen similar set ups in many posted home theater systems at AVSforum and others. When I demo such systems it just feels so mismatched. I'm watching the battle scene from Gladiator, I'm surrounded by what seems like a semi-life-sized battle, while watching these tiny figures running around on screen. Same for a sci fi movie, or star wars - I'm lifted off my feet by the gigantic space ship sounds, while watching toy-sized images of the space ships. My brain just has trouble mapping those together. The sound becomes distracting more than anything to me.

That's why I carefully chose speakers that would create a believable scale to the image size - like the very small Spendor S3/5 bookshelf speakers that flanked my 42" plasma for many years - perfect scale fit.

With the over 10 foot wide projection screen I could scale up the sound to match, so with the Hales speakers (and monitor audio surrounds) to me the scale fit is just perfect. It all feels believably like it's coming from the screen.

 

[MattHooper]

I can understand the potential benefits of adding a centre channel from an imaging/soudstanging perspective but that simulation does not bode well for the tonal accuracy side of things...

The best mapping of sound to image (with separate speakers) I've experienced have been my plasma flanked by Spendor S3/5 bookshelf speakers. I never needed a center channel because the image wasn't that wide and the phantom image just seemed to perfectly emanate from the screen. And this held up surprisingly well even off axis.

My projection screen was just too wide to pull this off. I did try just using my Hales monitors without a center channel and at the sweet spot only it mapped well to the screen. But I actually found the addition of the Hales center channel raised the timbral quality of the overall sound. Voices sound a bit "phasey" in that stereo imaging way with just the two channels. Whereas they sound more solid and rich with the center channel. Over all it enhanced the timbral realism of the whole system IMO. Which seems to be consistent with F. Toole's comments.

 

[phoenixdogfan]

The best mapping of sound to image (with separate speakers) I've experienced have been my plasma flanked by Spendor S3/5 bookshelf speakers. I never needed a center channel because the image wasn't that wide and the phantom image just seemed to perfectly emanate from the screen. And this held up surprisingly well even off axis.

My projection screen was just too wide to pull this off. I did try just using my Hales monitors without a center channel and at the sweet spot only it mapped well to the screen. But I actually found the addition of the Hales center channel raised the timbral quality of the overall sound. Voices sound a bit "phasey" in that stereo imaging way with just the two channels. Whereas they sound more solid and rich with the center channel. Over all it enhanced the timbral realism of the whole system IMO. Which seems to be consistent with F. Toole's comments.

That phantom image won't be any worse than the one enjoyed in two channel stereo reproduction by anyone seated in the sweet spot midway between the left and right speakers.

 

[Sancus]

but content often overlaps channels (e.g.- panning), and as the center is closer to the MLP than the left and right (~13" or ~1ms in my case), the effect for common frequencies could be similar to an early reflection. Or not?

Content only overlaps channels when you're trying to produce a phantom image, which is why phantom center images are not tonally correct. Toole covers this in his book.

Most center channel content that I've bothered to check is not duplicated in the L/R channels, or at least not significantly. Sometimes there is very low level reverb or room reflections type content but it's much lower in level than the direct sound in the center. And those aren't always the same in terms of frequency anyways.

If anything, proper use of a center channel substanially reduces interference effects because when mixed properly anything that would normally require L/R playing together only needs to be emitted by a single source.

And why stop at one? If I had a center channel speaker for my 120" screen, would phantom "center" not still be employed for sounds between L and C and between C and R?

Yes but that isn't usually done unless the audio source is actually moving from one place to another.

 

[Mnyb]

I’m slowly reading Tooles book , seeing this chapter might explain why I’m rather pleased with the presentation using trifield upmix for all content in my Meridian system sound is reasonable and relaxing in the whole sofa and as it said mono listening highlights the speakers own properties as I use 5 speakers all the time for the envelope, I don’t hear how daft my speakers really are the Meridian 5200 is “Ok” but not more pairing with sub + the matching center + trifield and some room DSP and tone control use makes it work . ( do I want speakers with modern acoustic properties or what )

 

[bjmsam]

That phantom image won't be any worse than the one enjoyed in two channel stereo reproduction by anyone seated in the sweet spot midway between the left and right speakers.

Per @Floyd Toole Figure 9.7:

Per ripple tank simulation:

If anything, proper use of a center channel substanially reduces interference effects because when mixed properly anything that would normally require L/R playing together only needs to be emitted by a single source.

That makes sense for content that is exclusive to the center (or any other single) channel.

Yes but that isn't usually done unless the audio source is actually moving from one place to another.

I did not realize that for two stationary sources, one between L and C, the other between C and R (say two people engaged in an interview), all dialogue is handled exclusively by the center channel speaker. With my phantom center setup, those sources are imaged clearly left of center and right of center (important on a large screen).

 

[Sancus]

I did not realize that for two stationary sources, one between L and C, the other between C and R (say two people engaged in an interview), all dialogue is handled exclusively by the center channel speaker. With my phantom center setup, those sources are imaged clearly left of center and right of center (important on a large screen).

Dunno what content you're talking about but the few times I've bothered to mute channels to actually see what they're playing, in a 3 channel setup I have literally never heard audio split equally between any two channels for film/tv content. 1 channel is almost always dominant and it's almost always the center channel except for multichannel music that is poorly mixed and using phantom center for no good reason. L or R can dominate occasionally if most audio is intended to be from that direction.

Film/TV content tends to rely on the ventriloquism effect and doesn't really bother trying to image positions across the screen purely using audio. Can't imagine it would do anything anyways since the ventriloquism effect eliminates imaging entirely in most cases. That effect also makes it hard to tell what is actually in each channel without muting the others.

Multichannel music varies much more than film/tv content however. But even so, phantom center is one of the most common effects in stereo audio and is known to produce incorrect tonality. Eliminating it means a big improvement no matter what other edge cases are introduced.

 

[MattHooper]

That phantom image won't be any worse than the one enjoyed in two channel stereo reproduction by anyone seated in the sweet spot midway between the left and right speakers.

Yup.

 

[hex168]

This looks interesting for a cheap coax center (DIY):

CX120-8 easy build - Techtalk Speaker Building, Audio, Video Discussion Forum

Since this is a full build thread, I am putting it in the forum rather than the gallery for now. https://www.parts-express.com/dayton-audio-cx120-8-4-coaxial-driver-with-3-4-silk-dome-tweeter-8-ohm--295-384 I have been asked a couple of times for a PEDS center channel. I made an MTM version...

techtalk.parts-express.com

 

[Cheyenne]

Hi all - Long time lurker - I appreciate the discussion and Erin and Amir's work on center channels. I have had and used a Polk CS200 MTM, an Energy Take 5 Classic TM, and currently have two set ups - one with an upright Dayton 442 (seems to work really well) and my main systen with Infinity Ref 252 and 152s. I use AVRs from Onkyo and Denon with Audyssey for set up.

One consistent thought I have had is why not a single full range driver like a BMR, Tang Bang 4 or 6" or Adire Audio driver. We consistently discuss issues with an MTM, claim the reason for a physical center is to focus dialog on the screen etc. Most dialog is in the 200-6.3K range according to Google so why aren't we focused on flat, wide dispersal in that range especailly when we know there is a sub for the bottom octave?

I am assuming timber matching is less of an issue with DSP/Audessey that at least matches levels.

Shouldn't we forget 20-20k and go with flat in the dialog range with wide dispersion from a single driver?

 

[More Dynamics Please]

Hi all - Long time lurker - I appreciate the discussion and Erin and Amir's work on center channels. I have had and used a Polk CS200 MTM, an Energy Take 5 Classic TM, and currently have two set ups - one with an upright Dayton 442 (seems to work really well) and my main systen with Infinity Ref 252 and 152s. I use AVRs from Onkyo and Denon with Audyssey for set up.

One consistent thought I have had is why not a single full range driver like a BMR, Tang Bang 4 or 6" or Adire Audio driver. We consistently discuss issues with an MTM, claim the reason for a physical center is to focus dialog on the screen etc. Most dialog is in the 200-6.3K range according to Google so why aren't we focused on flat, wide dispersal in that range especailly when we know there is a sub for the bottom octave?

I am assuming timber matching is less of an issue with DSP/Audessey that at least matches levels.

Shouldn't we forget 20-20k and go with flat in the dialog range with wide dispersion from a single driver?

If only dialog were sent to the center channel then a single driver might produce adequate performance. However, many sound effects are also sent to the center channel so those would be compromised without a full range center.

 

[Sancus]

Most dialog is in the 200-6.3K range according to Google so why aren't we focused on flat, wide dispersal in that range especailly when we know there is a sub for the bottom octave?

The center doesn't need to cover most dialog though, it needs to cover *all* possible dialog frequencies including overtones. The standard ranges for speech intelligibility go up to 8khz(which is why that's the max audiologists test for hearing), and down to 125hz. Unfortunately, even that is not enough, Barry White has a median vocal frequency of 71hz and it's normal, if less common, for men to have frequencies under 100hz. And then there's also voices altered by computer, which are quite common in many genres, that could be just about anywhere.

That's ignoring the foley issue mentioned by the poster above, where sound effects are also primarily in the center. In general, the Dolby/DTS/etc specs expect and treat the center as a full range speaker that is identical to the L/R. So anything else is out of spec and substandard.

 

[Cheyenne]

I appreciate the replies. I have tried to find more info on what exactly goes to the Center under DTS etc. I know they spec full range but what actually goes to the center?

This is all just musings for me since I use full range centers, but the above ripple tank sims made me start to think about it.

Thanks again for humoring the newb.

 

[Sancus]

I appreciate the replies. I have tried to find more info on what exactly goes to the Center under DTS etc. I know they spec full range but what actually goes to the center?

That isn't defined by spec, but rather is up to the director/audio engineers/etc. A good way to find out is to just play some movies/multichannel tv shows with L/R disconnected or turned off. You'll find that the center is basically always the loudest channel on average, by at least 3dB but sometimes much more.

The exception to this is multichannel music which varies a lot more and the center isn't always primary. But that's not relevant to home theatre really.

There are some examples in that thread.

 

[valerianf]

" So anything else is out of spec and substandard"
I fully disagree: in my experience, it is better to set the center channel to small and to let the large FL FR speakers handle
the low frequencies of the center channel.

In the similation made by Erin, there is a missing parameter.
The FL and FR speakers will create a phantom center speaker that will play at the same time as the physical center speaker.
What is very difficult is to get the same phase for these 2 center speakers.
To get it properly I had to modify the phase of the FL FR passive crossover.
When in phase the difference of sound quality is obvious: you get a uniform front sound stage.

 

[Sancus]

I fully disagree: in my experience, it is better to set the center channel to small and to let the large FL FR speakers handle
the low frequencies of the center channel.

In the similation made by Erin, there is a missing parameter.
The FL and FR speakers will create a phantom center speaker that will play at the same time as the physical center speaker.
What is very difficult is to get the same phase for these 2 center speakers.

Sorry, but there's no possibility for disagreement. The Dolby Atmos HT guidelines literally state, and I quote: "The frequency response of all speakers other than the subwoofer must conform to the wide-range characteristic defined in ISO 2969/SMPTE 202 standards, with or without bass management. The response must extend from 40 Hz at the low frequencies and ideally up to 18 kHz with no variation greater than ±3 dB." (pg15, 2.4.2)

If you disagree, take it up with the people who define the way HT audio is recorded and reproduced for effectively all modern content, not with me. It's got nothing to do with small/large at all. And anyway, Large is a useless setting for 99.99% of systems.

Also, I don't know what you mean by phantom center. There is no phantom center possible in much of the content because audio is not present in the L/R, like, at all. I just watched some DD5.1 material on Netflix and Amazon Prime and checked and a huge portion of the sound effects and dialog were center channel ONLY with at most a tiny amount of reverb in the L/R but usually nothing at all. Bass managing non-localizable frequencies does not change that. And that is a typical way this content is mixed.

 

[Doctors11]

Sorry, but there's no possibility for disagreement. The Dolby Atmos HT guidelines literally state, and I quote: "The frequency response of all speakers other than the subwoofer must conform to the wide-range characteristic defined in ISO 2969/SMPTE 202 standards, with or without bass management. The response must extend from 40 Hz at the low frequencies and ideally up to 18 kHz with no variation greater than ±3 dB." (pg15, 2.4.2)

If you disagree, take it up with the people who define the way HT audio is recorded and reproduced for effectively all modern content, not with me. It's got nothing to do with small/large at all. And anyway, Large is a useless setting for 99.99% of systems.

Also, I don't know what you mean by phantom center. There is no phantom center possible in much of the content because audio is not present in the L/R, like, at all. I just watched some DD5.1 material on Netflix and Amazon Prime and checked and a huge portion of the sound effects and dialog were center channel ONLY with at most a tiny amount of reverb in the L/R but usually nothing at all. Bass managing non-localizable frequencies does not change that. And that is a typical way this content is mixed.

When you "tell" your AVR that there is no center speaker, doesn't it take the center information and divide it between the L/R?

 

[bjmsam]

When you "tell" your AVR that there is no center speaker, doesn't it take the center information and divide it between the L/R?

Yes, resulting in comb filtering which can be audible and extremely annoying.

phantom center is one of the most common effects in stereo audio and is known to produce incorrect tonality. Eliminating it means a big improvement no matter what other edge cases are introduced.

You're right. I just finished building a center channel speaker (consistent with Erin's (@hardisj) video at 32:04; details here) and it eliminated my comb filtering problem! Panning still seems perfect, too.

in my experience, it is better to set the center channel to small and to let the large FL FR speakers handle
the low frequencies of the center channel.

You're right. My new center speaker exactly matches my left and right speakers down to 200Hz (identical drivers and crossover sans woofers) and sounds terrific; lower frequencies are non-directional so I can't tell that they are coming from left and right rather than center.

 

[valerianf]

f you disagree, take it up with the people who define the way HT audio is recorded and reproduced for effectively all modern content, not with me. It's got nothing to do with small/large at all. And anyway, Large is a useless setting for 99.99% of systems.

It has nothing to do with Atmos standard.
First I do not have Atmos but a 5.1 system.
Second my room is not a Atmos qualified room but an everyday living room.
Also I am able to play with the adjustment of the Avr and notice when the sound quality is improving.
This is called system tuning and has nothing to do with the Atmos standard.

My car is not standard from manufacturer but modified to get some workaround about the design issues.
Every user has different requirements.