Are three channels better than two for stereo reproduction?

[gnarly]

Interesting setup. If I may ask, what is the hardware and how did you configure your remote to switch from LCR matrixing to stereo?

Hardware is either a Q-Sys Core 110f, or 510i, defending on how much FIR processing I'm using.
With Q-Sys, you design the processing schematic on a PC, which loads the schematic it into the hardware Core.
Any control control component, adjustment, etc in the schematic can be put built into a custom remote that runs on PC or IOS devices.
Here's the emote I'm using now on a touchscreen laptop.

 

[Duke]

I believe in their [Magnepan's] setup, the central speakers were used unidirectionally with the back radiation absorbed... while the L&R were used normally is dipole/bipole speakers further out from reflecting surfaces

That makes a lot more sense to me! Of course the attenuation of the backwave would be reduced at longer wavelengths, but ime it is mostly in the midrange and treble region where too-early-arrival of the backwave energy is detrimental, and I'm sure Magnepan took the time to get it right.

 

[gnarly]

Great information!

The paper to which you provided the link derives "The New Optimal Matrix" which, according to the paper, maximizes signal separations. I am curious if you gave that one a try. If so, what differences did you detect between that one and the Gerzon matrix you are using?


View attachment 382364

Thx!



Yes, the EP optimal is actually my favorite. I didn't want to mention it, to keep from complicating my example beyond the Gerzon paper.
The EP optimal matrix is the Exp button on the remote shown in #61.

Gerzon1 is what I showed earlier, and MS matrix is the better known standard.

I like the EP optimal best because Gerzon1 puts too little into center, and MS too much.
It of course varies by track, but EP does seem to be optimal

What I really need to do next, and what Gerzon did, is apply a high-pass to the matrix, that leaves lower freq's alone.
Next on the experiment list.

 

[terryforsythe]

What matters most to me is what sounds best/most realistic.

I fully agree.

I have never played with a center channel, and it does sound intriguing, especially after reading the paper to which @gnarly posted the link, as well as some other papers I read after reading that one.

That being said, if cost is an issue, I would start with a stereo setup using speakers that sound stage and image very well, then play with their room positioning and toe. Doing so, my systems sound stage and image extremely well with solid, well defined phantom images in the sound stage. Although the combing effect may be present, with its dilution due to room reflections, I have never noticed the change in tonal response (I may now if I listen for it, though, after going down the rabbit hole). Nonetheless, the net change in tonal response that may be present does not seem to degrade the definition and solidity of the phantom images, at least in my systems/rooms.

If you are not able to achieve satisfactory phantom images using two speakers with good sound stage and imaging, you always can add a third with the appropriate electronics.

 

[LowShelfFilter]

If you are not able to achieve satisfactory phantom images using two speakers with good sound stage and imaging, you always can add a third with the appropriate electronics.

For clarification, I think the phantom imaging is pretty good in my system, but I can only say this in reference to my own system. Without having heard a good 3-channel system, I can't really know if my system's phantom images are that good. But I wouldn't say my system "transports" me to the performance or anything like that. And when I was visualizing potential improvements, a center channel came to mind.

This is the main quote from Toole's book on this topic which stuck with me:
“The stereo phantom center will suffer from a significant dip in the spectrum around 2 kHz because the sound from both loudspeakers reaches both ears at different times. This interaural crosstalk cancellation is the first dip in an acoustical interference comb filter as explained in Figure 7.2. The consequence of this is that the sound quality of the phantom center image (usually the featured artist) in stereo recordings is fundamentally compromised. Those who believe that phase shift is audible and that waveform fidelity is essential must think again, because for this sound image both are seriously corrupted.”

Excerpt From
Sound Reproduction
Toole, Floyd E.;
This material may be protected by copyright.

 

[terryforsythe]

For clarification, I think the phantom imaging is pretty good in my system, but I can only say this in reference to my own system. Without having heard a good 3-channel system, I can't really know if my system's phantom images are that good. But I wouldn't say my system "transports" me to the performance or anything like that. And when I was visualizing potential improvements, a center channel came to mind.

This is the main quote from Toole's book on this topic which stuck with me:
“The stereo phantom center will suffer from a significant dip in the spectrum around 2 kHz because the sound from both loudspeakers reaches both ears at different times. This interaural crosstalk cancellation is the first dip in an acoustical interference comb filter as explained in Figure 7.2. The consequence of this is that the sound quality of the phantom center image (usually the featured artist) in stereo recordings is fundamentally compromised. Those who believe that phase shift is audible and that waveform fidelity is essential must think again, because for this sound image both are seriously corrupted.”

Excerpt From
Sound Reproduction
Toole, Floyd E.;
This material may be protected by copyright.

What speakers are you using?

Have you tried DSP to tune the speakers to achieve the same frequency response measurements at the listening position?

When Toole states the phantom center image is compromised, my understanding from reading that chapter is that he is discussing the tonal balance, not necessarily the solidity of the phantom image. Also, the level of tonal change is worse case in an anechoic chamber, but less of an issue in a typical room due to room reflections. My understanding is reinforced by my own experience with my systems.

I have had speakers that were not stellar at providing rock solid phantom images, most notable a pair of paradigm speakers I had mounted on the wall on either side of my TV. They were not good at all. But, my current speakers sound stage and image very well, and the phantom images are solid.

 

[LowShelfFilter]

What speakers are you using?

Neumann KH120ii

Have you tried DSP to tune the speakers to achieve the same frequency response measurements at the listening position?

I have them symmetrically positioned relative to the listening position and only DSP tune the summed response below my room's transition frequency, since the speakers are already flat on-axis.

When Toole states the phantom center image is compromised, my understanding from reading that chapter is that he is discussing the tonal balance, not necessarily the solidity of the phantom image. Also, the level of tonal change is worse case in an anechoic chamber, but less of an issue in a typical room due to room reflections. My understanding is reinforced by my own experience with my systems.

“Figure 7.1 illustrates the basic sets of reflections for the loudspeakers in stereo and the front channels in home theaters. It is evident that the sounds arriving at the listener are dependent on the loudspeaker directivity patterns and the acoustical performance of the reflecting points on the room boundaries. In general, for forward-firing loudspeakers, sounds radiated to the rear are mainly low frequencies. For this reason they are not a significant factor, but for some designs (bipoles, dipoles, omnidirectional) they are. Apparent in this illustration is the dramatic simplification in the reflection pattern of a real center channel loudspeaker compared to the “double-mono” phantom center image. They cannot sound the same in terms of timbre or space, not even considering the interaural crosstalk dip at 2 kHz. The importance of geometric symmetry in stereo setups is clear, including a “sweet spot” seat, if one wishes to hear the recording as it was created.”

Excerpt From
Sound Reproduction
Toole, Floyd E.;
This material may be protected by copyright.

 

[terryforsythe]

Neumann KH120ii

You have good speakers. They should be capable of good imaging as long as the speakers are positioned so that the tweeters are level with ear height, within +/- 10 deg. or so.

I have them symmetrically positioned relative to the listening position and only DSP tune the summed response below my room's transition frequency, since the speakers are already flat on-axis.

I would measure their frequency response at the listening position with REW. Even though the speakers are flat on-axis, room reflections may adversely affect the frequency response at the listening position. For me, this was more of an issue in my family room, where the speakers are over 3m from my listening position and the room is asymmetric. In my office my speakers are less than 2m from my listening position, and DSP optimization doesn't have much impact on imaging.

They cannot sound the same in terms of timbre or space, not even considering the interaural crosstalk dip at 2 kHz.

I agree they will not sound the same, and that there will be a difference in terms of timbre and space, but does that difference mean that you cannot obtain good sound staging with solid images using two speakers? My experience with my systems tells me no. Both of my systems sound stage very well and produce solid images using two speakers. The posts in this thread by @tmtomh and @Hear Here provide similar experiences.

Anecdotally, a few months back, after tuning my family room system, I had some friends/family members sit in my listening chair, close their eyes, and listen. They were blown away that the music was being produced by only two speakers (and a subwoofer). More than one person noted that it sounded like the singer was actually standing in the middle of the room.

As stated by Toole, "draw your own conclusions. It may be that 'one size does not fit all.'" The impact of the room on the sound is significant, and every room is different.

I have never played with a center channel, and perhaps it will improve sound staging and imaging even further. I don't know, but I do find the subject intriguing. If the cost of another speaker and electronics to implement that are not an issue for you, perhaps give it a try. @gnarly seems to like it.

 

[LowShelfFilter]

Even though the speakers are flat on-axis, room reflections may adversely affect the frequency response at the listening position.

The on-axis sound is said to be perceptually dominant above the room transition frequency due to the precedence effect, although I am sure you already know this.

Here is a spatial average of my summed response (3x3 measurement centered at listening position):

I have only done equalization below 250 Hz. I experimented with removing the 300 Hz resonance and it subjectively worsened sound quality. My guess for why this is, is that it is caused by rear-wall reflections since the back of my head is very close to the rear wall.

If I limit the spatial average to the 6 points not directly next to the wall, it looks more normal

 

[terryforsythe]

Here is a spatial average of my summed response (3x3 measurement centered at listening position):

Can you post a graph showing the left and right channels individually? I'm looking to see if there are any significant differences between the two channels above 400 Hz.

Your bass response looks very, very good!

 

[LowShelfFilter]

Can you post a graph showing the left and right channels individually? I'm looking to see if there are any significant differences between the two channels above 400 Hz.

Your bass response looks very, very good!

I do not have such a measurement of my current calibration, but here is a moving microphone method measurement from when I mistakenly calibrated bass using the individual channel responses. Equalization was only performed below 500 Hz here.

Also, just to reiterate, I would say the phantom imaging of my system is pretty good, but it's obvious that there is not a real sound object emanating from in front of me. The phantom center has this, not fully natural, stretched presentation. Like, I can vaguely tell the singer is somewhere in the center, but I can't pinpoint exactly where as if there were a singer actually in front of me.

 

[terryforsythe]

I do not have such a measurement of my current calibration, but here is a moving microphone method measurement from when I mistakenly calibrated bass using the individual channel responses. Equalization was only performed below 500 Hz here.

The responses look very close to one another. Overall, I think it looks very good. I'm not optimistic that doing any further DSP tuning will be of benefit to your imaging.

the back of my head is very close to the rear wall.

I am wondering whether this has an adverse impact on imaging. If you have adequate room to do so, as an experiment, you may try sitting further forward, away from the back wall a few feet or so, to see if it makes any difference. If it does, at least you found the culprit.

 

[Koeitje]

Generally speaking? If you have a way to do a stereo to LCR encode, then yes!

Stereo's biggest weakness is that the phantom center is never particularly strong.

That's not how I would put it. I would say that a phantom center is too reliant on the listening position, but if you can sit in the middle there is no need for an actual center. If you can sit in the sweet spot the only thing a center brings to the table is more SPL (and thus lower distortion overall).

 

[LowShelfFilter]

I am wondering whether this has an adverse impact on imaging. If you have adequate room to do so, as an experiment, you may try sitting further forward, away from the back wall a few feet or so, to see if it makes any difference. If it does, at least you found the culprit.

The backwall reflections positively contributed to a sense of 3D sound (almost like rear speakers), but it was damaging forward imaging a bit too much for my taste. After putting two 24"x48"x5.25" fiberglass acoustic panels behind me, this seemed to help. Going forward damages bass response a lot, and my speakers are positioned such that they point at my ears when sitting at the listening position, but the sound does become "brighter" as I move closer to the speakers, but this is expected since sound gets louder as I get closer to the source and the tonal balance is becoming more high frequency biased. I would say that overall imaging becomes less "realistic" the closer I get to the speakers. The closer I get, holes start appearing in the phantom center, and the sweet spot becomes much smaller.

Also, here is RT60 measured at listening position, for both channels:

 

[goat76]

Also, just to reiterate, I would say the phantom imaging of my system is pretty good, but it's obvious that there is not a real sound object emanating from in front of me. The phantom center has this, not fully natural, stretched presentation.

My phantom center doesn't have a stretched presentation as you describe. The snare drum, kick drum, and the woman's voice (in the sound file I posted earlier) are equally as wide wherever they are panned in the stereo field. I suggest you use the recording and see if you can decrease the distance between your loudspeakers while also making toe-in adjustments, and at the same time keep the listening position to an equilateral triangle until you hear the same width of all the sounds in the recording no matter where they are panned in the stereo field. That's a good way to determine what distances work for your loudspeakers.

 

[LowShelfFilter]

My phantom center doesn't have a stretched presentation as you describe. The snare drum, kick drum, and the woman's voice (in the sound file I posted earlier) are equally as wide wherever they are panned in the stereo field. I suggest you use the recording and see if you can decrease the distance between your loudspeakers while also making toe-in adjustments, and at the same time keep the listening position to an equilateral triangle until you hear the same width of all the sounds in the recording no matter where they are panned in the stereo field. That's a good way to determine what distances work for your loudspeakers.

The fully left and fully right parts are the clearest.

 

[terryforsythe]

The fully left and fully right parts are the clearest.

I'm leaning toward your room, or the speakers' placement in the room, as being the issue. You could experiment with different speaker positions and orientations (i.e., not pointed directly at you), and/or delve into room treatment. Absent that, it may be worthwhile to give a shot at adding a center speaker. If you do so, please keep us updated. I'm curious how it turns out.

 

[LowShelfFilter]

Are we sure this is an issue with my setup and not physical limitations of having two speakers? I feel like it is very difficult to have a conversation about how my system sounds and what improvements can be made without having the people giving advice also listen to the system. The reason I bring this up is because even if I move closer to the speakers (which lowers rear wall reflections), the image is still more stretched when not hard panned and the sweet spot becomes way smaller. I find it hard to believe that with just speaker positioning and acoustical treatment, I can have phantom images be as strong as real images, especially considering the acoustical interference at play with phantom images.

 

[MaxwellsEq]

but is fidelity with regard to an inherently compromised medium worth preserving.

The whole point of an audio playback system is to mess with the original source as little as possible. You have two data sets - a HiFi system should preserve this data, not corrupt it with a 3rd, made-up data set.

You started with this observation:

Considering that two speakers cannot accurately reproduce a 3D acoustic event in the first place

Which suggests you have not heard it done properly. You are trying to solve a problem which doesn't exist with properly setup systems.

What matters most to me is what sounds best/most realistic

Actually, it appears that it does not matter to you what is most realistic, since you are "creating" a sound generator which doesn't exist in the source material.

this method would actually result in a more realistic sound presentation (don't know if it actually will), then I would be fine with pursuing it.

It can't be more realistic, only less realistic.

You might prefer it, but that's a completely different matter.

 

[LowShelfFilter]

The whole point of an audio playback system is to mess with the original source as little as possible. You have two data sets - a HiFi system should preserve this data, not corrupt it with a 3rd, made-up data set.

You started with this observation:

Which suggests you have not heard it done properly. You are trying to solve a problem which doesn't exist with properly setup systems.

Actually, it appears that it does not matter to you what is most realistic, since you are "creating" a sound generator which doesn't exist in the source material.

It can't be more realistic, only less realistic.

You might prefer it, but that's a completely different matter.

Out of curiosity, do you think an anechoic chamber is the ideal listening environment? I used to focus more on fidelity with respect to the recording, but I eventually realized that stereo recordings are 2-dimensional, while real sound is 3-dimensional. I am trying to achieve fidelity with respect to real sound, not a 2-dimensional recording.

Why do audio legends like Floyd Toole up-mix if stereo is so good and realistic?