Are three channels better than two for stereo reproduction?

[dlaloum]

My experience has been that even with a high-end trinaural processor, the depth of the center of the soundstage is constrained relative to normal two-channel, and I have customers who have gone from trinaural back to two-channel because this was their experience as well. However this may be system-dependent; I think it's quite possible that in some systems there would be no significant difference in the soundstage depth between two-channel and upmixed three-channel.

Not my experience, and not what was reported in the article on the Magneplanar three channel setup I linked above...

 

[Trdat]

The only time a discrete center channel matters is for dialog, and it must be discrete. Must have been recorded fo4 that purpose.

I agree, I have centre channel for watching YouTube and television, works great even though it took me a while to understand and absorb the benefits. However, you need a proper upmixer such as dolby movie for this.

I have experimented with 5.1 and centre channels for music and I have never felt that it brings any benefits to the table. I remember when I went through the upmixing/surround phase during the early days of being on ASR and wanting to experiment and some would say just stick to stereo now I understand what they were getting at.

Lowshelffilter, I know you are specifically talking about adding a centre but at least the consensus from the pages of threads i went through was this,

1. Use surround or 5.1 or centre channel for music that is designed for that purpose if your listening to stereo recorded pieces than stick to stereo.
2. Only Classical or music similar to this might have some benefit in upmixing but then again your better off choosing recordings that are discretely mixed for 5.1 otherwise stick to stereo.

Good luck with your endevour

Edit. This is not to say that many people on this forum are not fans of upmixing or a centre channel for music, its just many of these people are using discrete recordings to enjoy the centre or 5.1 or with specific types of music that work well with it.

 

[Duke]

Not my experience, and not what was reported in the article on the Magneplanar three channel setup I linked above...

If I understand that article correctly, the system they used consisted of left and right wall-mounted dipole speakers (the reflecting wall being immediately behind the dipoles) and a center-channel dipole that was likewise very close to the wall behind it. I understand the reason for placement so close to the front wall, but that's not something I'd want to do with dipoles, as ime it degrades the perception of soundstage depth relative to a setup having significantly more distance to the wall behind the speakers.

The trinaural system I heard also used bi-directional speakers, but they were positioned six or eight feet from the front wall. So I think the two setups were different enough in how they interacted with the wall behind them that conclusions drawn from one might not necessarily be applicable to the other.

 

[MaxwellsEq]

Using a MiniDSP 2x4 product, you can route audio as such:
Output 1: Right Speaker (Right input channel)
Output 2: Left Speaker (Left input channel)
Output 3: Center Speaker (Right + Left input channels)
Output 4: Subwoofer(s) (Right + Left input channels)

This is what I am considering doing.

You didn't understand. Published stereo content only has TWO sets of data for you to work with. You can NOT create a 3rd centre channel.

What you can do is mashup the two sets of data you have. BUT, since you have no data about the microphones' physical layout or how panning was used on the mixer, you are guessing. You are not producing a genuine "centre channel", but your own personal opinion on what should be there. This is not fidelity (as in HiFi - High Fidelity).

 

[dlaloum]

If I understand that article correctly, the system they used consisted of left and right wall-mounted dipole speakers (the reflecting wall being immediately behind the dipoles) and a center-channel dipole that was likewise very close to the wall behind it. I understand the reason for placement so close to the front wall, but that's not something I'd want to do with dipoles, as ime it degrades the perception of soundstage depth relative to a setup having significantly more distance to the wall behind the speakers.

The trinaural system I heard also used bi-directional speakers, but they were positioned six or eight feet from the front wall. So I think the two setups were different enough in how they interacted with the wall behind them that conclusions drawn from one might not necessarily be applicable to the other.

I agree that the illusion of solidity is tricky and very setup dependent ... and there are probably hundreds of solutions to the problem!

In the magnepan case the inside l/c/r speakers were all fed the same signal - making a multi driver speaker - and then further out, the full range panels were the true L/R.

But they also messed with the PLII configuration parameters - and they always kept the parameters they used for this exercise secret.... (sadly)

I believe in their 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

 

[goat76]

Considering that two speakers cannot accurately reproduce a 3D acoustic event in the first place, would using a center channel summed to mono with LR playing normally provide for a superior/more realistic listening experience than just using LR? Please note that I am not referring to listening to three channel recordings, but rather up-mixing two channel recordings to three channels.

From Floyd Toole's book, there are significant acoustic advantages to using a center channel. Are there any downsides to using LCR for stereo reproduction instead of LR?

I don't think adding a center speaker will make a difference to how the three-dimensional space in the recording will be perceived, the particular stereo fault that is described is more of a problem that affects the tonality of phantom-centered sounds.

Maybe you just need to optimize your speaker's position, the listening position, and your room acoustics to reach your goal. In my opinion, the combination of a high ratio of direct sound and optimization of the speakers to get a solid phantom center are the most important things that will give the best insight into the three-dimensional space of the recording. My suggestion is that you shrink down the listening triangle and use acoustic treatment to increase the direct sound from your speakers, and optimize the distance between your speakers so they together create a unified stereo image with a solid phantom center.

You can use the following sound file I made, it contains a snare drum, a kick drum, and a voice that pan all the way from the left to the right. They all should sound equally as solid in all the positions if the speaker positioning is correct.

Dropbox

www.dropbox.com

 

[EJ3]

I agree that the illusion of solidity is tricky and very setup dependent ... and there are probably hundreds of solutions to the problem!

In the magnepan case the inside l/c/r speakers were all fed the same signal - making a multi driver speaker - and then further out, the full range panels were the true L/R.

But they also messed with the PLII configuration parameters - and they always kept the parameters they used for this exercise secret.... (sadly)

I believe in their 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

The setup seems to have been specifically for the (at the time) the Mandalay Hotel rooms. (Please correct me if I am wrong, my wife tells me that I have been wrong before).

 

[Hear Here]

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

That's not the case if speakers are well chosen and properly set up.

Certainly, the wrong speakers, or probably more generally, the wrong TYPE of speakers, will offer poor imaging and one may be tempted to try a centre channel, but the right speakers, well set up have no need for a centre speaker.

I've always been impressed with the imaging of my own horn speakers, but was thinking that omnis should be the best type of speaker for my particular room. I visited showrooms with demos of the 2 leading omni brands, but neither offered anything like the imaging my horns offered. If one closes one's eyes and points a towards particular instrument or singer, how confident are you with the accuracy of this test? With the omnis, there was a vagueness I never experience with the horns. A friend has big Quad panels and his system perhaps offers even better imaging. He has a centre speaker (as it's part of his HT system) and I was convinced this centre speakers as working - but not so.

One rarely hears a really well set up system but it is startling when one does. The disadvantage of course is the very small sweet spot, but I live with that. No centre channel needed in my own system!

 

[Petrushka]

That's not the case if speakers are well chosen and properly set up.

Certainly, the wrong speakers, or probably more generally, the wrong TYPE of speakers, will offer poor imaging and one may be tempted to try a centre channel, but the right speakers, well set up have no need for a centre speaker.

I've always been impressed with the imaging of my own horn speakers, but was thinking that omnis should be the best type of speaker for my particular room. I visited showrooms with demos of the 2 leading omni brands, but neither offered anything like the imaging my horns offered. If one closes one's eyes and points a towards particular instrument or singer, how confident are you with the accuracy of this test? With the omnis, there was a vagueness I never experience with the horns. A friend has big Quad panels and his system perhaps offers even better imaging. He has a centre speaker (as it's part of his HT system) and I was convinced this centre speakers as working - but not so.

One rarely hears a really well set up system but it is startling when one does. The disadvantage of course is the very small sweet spot, but I live with that. No centre channel needed in my own system!

I suggested on another thread that horn speakers have an extended near field effect. Which makes them seductive even when their frequency response isn’t flat.

I’ve only heard one system that gave me the sustained illusion of being there, and that was an XD theater.

What’s remarkable is it’s the only movie theater I’ve been in that was not too loud.

 

[EJ3]

I suggested on another thread that horn speakers have an extended near field effect. Which makes them seductive even when their frequency response isn’t flat.

I’ve only heard one system that gave me the sustained illusion of being there, and that was an XD theater.

What’s remarkable is it’s the only movie theater I’ve been in that was not too loud.

I was in a movie theatre yesterday that I thought was too loud and my wife had her fingers in her ears, so I was not the only one. But that was in the adds leading up to the movie "The Never Ending Story" (which I & my wife had never seen before).
When the movie came on, the sound was at a well tolerated leveI (and in a few spots: too low).
I guess that when the movie originally came out, I was a 2 busy single 27 year old and today, for me, it was a "just OK" movie but my wife (who is 7 years younger than myself) asked me to never take her to anything like that again.
What she got out of it is that it promoted children to cut school to not actually deal with their problems or learn anything and waste their day fantasizing.
We have a 36 year old son that is an English teacher and soccer coach, so whatever she did raising him (I had very little to do with it, as my job had me somewhere else in the world, usually) seems to have worked pretty well.
So I will defer to her opinion, as I did not get much from this movie.
(perhaps because I like to read very much and I don't need encouragement to read and visualize/fantasize (which is what seemed to be the theme of the movie: read and use your imagination).

 

[JSmith]

What she got out of it is that it promoted children to cut school to not actually deal with their problems or learn anything and waste their day fantasizing.

I must say, no offence intended, but that is the strangest opinion of the movie I've ever heard. Developing and exploring ones imagination is key to creativity. The fact he ended up cutting school was incidental to the adventure and the lessons learned.


JSmith

 

[Petrushka]

I must say, no offence intended, but that is the strangest opinion of the movie I've ever heard. Developing and exploring ones imagination is key to creativity. The fact he ended up cutting school was incidental to the adventure and the lessons learned.


JSmith

Cutting school is a common fantasy.

 

[dlaloum]

The setup seems to have been specifically for the (at the time) the Mandalay Hotel rooms. (Please correct me if I am wrong, my wife tells me that I have been wrong before).

I thought they had this experimental setup at their own premises ... that was my impression from the article?

 

[EJ3]

I must say, no offence intended, but that is the strangest opinion of the movie I've ever heard. Developing and exploring ones imagination is key to creativity. The fact he ended up cutting school was incidental to the adventure and the lessons learned.


JSmith

I agree, it totally threw me for a loop.
My wife is Chinese, having grown up in mainland China, where education was revered above all else.
A very different perspective on regimented education.

 

[gnarly]

http://decoy.iki.fi/dsound/ambisonic/motherlode/source/7038.pdf

Math involved. Michael Gerzon paper from 1991. Optimally you need to do some frequency contouring, and some mid-side processing. Simplifying that paper some, one obvious result is plopping a center channel down and connecting it with mono combination of L and R will cause a narrow stereo width. A simple way to help with that is to widen the spacing between L and R to at least 90 degrees as viewed from the listeners position. You see lots of people saying just connecting a center narrows the stereo effect.

So simple terms widen your speakers unless you already have something of a hole in the middle. Reduced the center channel vs side channels by 3 to 6 db. Have the center equi-distant from the listener vs side speakers. Another aspect is some recording methods are effected differently than others doing a 2 channel to 3 speaker setup. So it might be better for some, a little worse for others, but without the extra matrix processing etc it won't always be for the better.

Yes. Here's a link from a fellow audio enthusiast that helped me implement Gerzon's matrices for turning stereo into 3 channel L-C-R.

http://elias.altervista.org/html/3_speaker_matrix.html

I'm currently running this matrix, using a wider 110 degree Left to Right arc like you suggest.

Just a few personal observations from comparing this kind of energy preserving LCR matrixing vs stereo.....

First, what I've found so far to be requirements for best results:
The three speakers need to be identical. Unfortunately, identical doesn't just mean same model. It means they would measure the same.
Variance between speakers diminish both stereo and LCR ime.
Way to check for speakers being same is play mono on both L& R sides. Should get a near rock solid phantom center. If not, all bet's are off until fixed ime.

Like already pointed out, all 3 speakers need to be equidistant on an arc to listener. This can be difficult in rooms, where the center speaker is likely to be closer than L & R.
In that case I've found it essential to both delay the center speaker, as well as attenuate it down to match L&R. (With an energy preserving matrix ala Gerzon, there is no need to reduce the center channel past what the matrix does.)
Ime, Sound field depth comparisons, stereo vs LCR, don't work without the speakers being either equidistant, or center's delay and attenuation being spot on.

With those prerequisites accomplished, I really like L-C-R.
Doesn't help all stereo tracks by any means, but I'd say 80% are improved. (I keep the ability to switch between them on the remote)
Center image, whether vocal or instrument gets stronger / more defined. And doesn't vary so much when moving off of the central axis. In fact, sound stays good within at least 1/3 the width of L & R.
It does shrink the width of the soundstage a little...perhaps the only real drawback. Most of the time, going back to stereo to regain the width, makes the original width sound bloated and loose. But not always....this falls within the estimated group of 20% of tracks that don't improve with LCR.
Sometime a recording just needs the width back...usually ones with a lot of reverb etc in them. Usually the kind I don't like lol.
We all have our preferences, eh?

 

[hex168]

Yes. Here's a link from a fellow audio enthusiast that helped me implement Gerzon's matrices for turning stereo into 3 channel L-C-R.

http://elias.altervista.org/html/3_speaker_matrix.html

I'm currently running this matrix, using a wider 110 degree Left to Right arc like you suggest.

View attachment 382361


Just a few personal observations from comparing this kind of energy preserving LCR matrixing vs stereo.....

First, what I've found so far to be requirements for best results:
The three speakers need to be identical. Unfortunately, identical doesn't just mean same model. It means they would measure the same.
Variance between speakers diminish both stereo and LCR ime.
Way to check for speakers being same is play mono on both L& R sides. Should get a near rock solid phantom center. If not, all bet's are off until fixed ime.

Like already pointed out, all 3 speakers need to be equidistant on an arc to listener. This can be difficult in rooms, where the center speaker is likely to be closer than L & R.
In that case I've found it essential to both delay the center speaker, as well as attenuate it down to match L&R. (With an energy preserving matrix ala Gerzon, there is no need to reduce the center channel past what the matrix does.)
Ime, Sound field depth comparisons, stereo vs LCR, don't work without the speakers being either equidistant, or center's delay and attenuation being spot on.

With those prerequisites accomplished, I really like L-C-R.
Doesn't help all stereo tracks by any means, but I'd say 80% are improved. (I keep the ability to switch between them on the remote)
Center image, whether vocal or instrument gets stronger / more defined. And doesn't vary so much when moving off of the central axis. In fact, sound stays good within at least 1/3 the width of L & R.
It does shrink the width of the soundstage a little...perhaps the only real drawback. Most of the time, going back to stereo to regain the width, makes the original width sound bloated and loose. But not always....this falls within the estimated group of 20% of tracks that don't improve with LCR.
Sometime a recording just needs the width back...usually ones with a lot of reverb etc in them. Usually the kind I don't like lol.
We all have our preferences, eh?

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

 

[terryforsythe]

Yes. Here's a link from a fellow audio enthusiast that helped me implement Gerzon's matrices for turning stereo into 3 channel L-C-R.

http://elias.altervista.org/html/3_speaker_matrix.html

I'm currently running this matrix, using a wider 110 degree Left to Right arc like you suggest.

View attachment 382361


Just a few personal observations from comparing this kind of energy preserving LCR matrixing vs stereo.....

First, what I've found so far to be requirements for best results:
The three speakers need to be identical. Unfortunately, identical doesn't just mean same model. It means they would measure the same.
Variance between speakers diminish both stereo and LCR ime.
Way to check for speakers being same is play mono on both L& R sides. Should get a near rock solid phantom center. If not, all bet's are off until fixed ime.

Like already pointed out, all 3 speakers need to be equidistant on an arc to listener. This can be difficult in rooms, where the center speaker is likely to be closer than L & R.
In that case I've found it essential to both delay the center speaker, as well as attenuate it down to match L&R. (With an energy preserving matrix ala Gerzon, there is no need to reduce the center channel past what the matrix does.)
Ime, Sound field depth comparisons, stereo vs LCR, don't work without the speakers being either equidistant, or center's delay and attenuation being spot on.

With those prerequisites accomplished, I really like L-C-R.
Doesn't help all stereo tracks by any means, but I'd say 80% are improved. (I keep the ability to switch between them on the remote)
Center image, whether vocal or instrument gets stronger / more defined. And doesn't vary so much when moving off of the central axis. In fact, sound stays good within at least 1/3 the width of L & R.
It does shrink the width of the soundstage a little...perhaps the only real drawback. Most of the time, going back to stereo to regain the width, makes the original width sound bloated and loose. But not always....this falls within the estimated group of 20% of tracks that don't improve with LCR.
Sometime a recording just needs the width back...usually ones with a lot of reverb etc in them. Usually the kind I don't like lol.
We all have our preferences, eh?

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?

 

[terryforsythe]

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

This probably could be setup using any DSP with at least 3 output channels having adjustable gain, wherein at least one of the output channels can be a combination of two of the input channels (so long as the channel is not limited in frequency, like a subwoofer out).

miniDSP has a number of units that would work, e.g., 2x4HD, Flex, SHD, etc.

CamillaDSP running an a Raspberry Pi would be a very cost effective solution.

 

[tmtomh]

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.

Per @terryforsythe 's comment above, are you talking about a phantom center that remains stable and/or as strong as L and R across a wide range of positions in the room? I ask because I've had several setups with very strong phantom centers, and from the main listening position my current setup has a fantastic phantom center that I perceive to be as strong as the L and R.

To be clear, I have no interest in arguing against a 3-channel setup for stereo. My father used to tell me about that (he saw and heard some demos in the late 1950s/early '60s that really made an impression on him). My only personal concern about 3-channel is that one thing I like about the phantom center is that it's phantom - the illusion of realism, or just the enjoyable impression of some kind of presence regardless of realism, is enhanced for me personally because I can't even begin to visually localize it to any speaker.

 

[LowShelfFilter]

You didn't understand. Published stereo content only has TWO sets of data for you to work with. You can NOT create a 3rd centre channel.

What you can do is mashup the two sets of data you have. BUT, since you have no data about the microphones' physical layout or how panning was used on the mixer, you are guessing. You are not producing a genuine "centre channel", but your own personal opinion on what should be there. This is not fidelity (as in HiFi - High Fidelity).

I know my proposed audio routing is not high fidelity with regard to the source file, but is fidelity with regard to an inherently compromised medium worth preserving? What matters most to me is what sounds best/most realistic. If 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.