Properties of speakers that creates a large and precise soundstage

[Axo1989]

If you take a stereo recording and remove the side channels, all that's left is THE mono signal.

You may be talking about M-S, which has one "side" channel. If you remove both "side" channels of L-R stereo, you'll get silence.

 

[LevityProject]

The sides, or side channel is always stereo, so it's often called "Sides" for short. I can refer to it just as "the side channel" if that helps.

But yes, it is one stereo channel and Removing that channel leaves just the mono signal ... Which means it's a fact that the side channel is what makes stereo ... Stereo, whether playing it back or recording it.

Manipulating the side channel with DSP is mid-side processing. Manipulating it in a speaker set up is where room reflections, directionality, speaker placement, and the listening position all come into play. It's non electric or DSP related mid-side manipulation for lack of a better term.

But It's not as if the mid and side channels only exist with encoding to reveal the separate channels. Mid and side pre-exist in a stereo field, otherwise there would be nothing to encode for DSP to manipulate it.

 

[Axo1989]

The sides, or side channel is always stereo, so it's often called "Sides" for short. I can refer to it just as "the side channel" if that helps.

But yes, it is one stereo channel and Removing that channel leaves just the mono signal ... Which means it's a fact that the side channel is what makes stereo ... Stereo, whether playing it back or recording it.

Manipulating the side channel with DSP is mid-side processing. Manipulating it in a speaker set up is where room reflections, directionality, speaker placement, and the listening position all come into play. It's non electric or DSP related mid-side manipulation for lack of a better term.

But It's not as if the mid and side channels only exist with encoding to reveal the separate channels. Mid and side pre-exist in a stereo field, otherwise there would be nothing to encode for DSP to manipulate it.

Ahh so that's where you are getting lost: "one stereo channel". Think about that non sequitur for a moment. You've apparently talked yourself into believing the side channel is stereo (calling it "side channel" or "sides" makes no difference) all by itself. It isn't.

And you don't require DSP to record or mix as M-S, it can all be done in the analog domain. Either way, when you listen via speakers/headphones, M+S=L and M+(-S)=R. You need both M and S to make stereo (M without S is mono, S without M is also mono). You can't make stereo out of S.

 

[goat76]

I'm not sure what you two are discussing, but I just want to point out that when it comes to mixing of a Mid/Side recording, it's no longer just one "Side" channel, it's actually two of them. So in Mid/Side mixing, there are three separate tracks in total, one Mid channel and two Side channels.

The recording of the "Sides" made with the Figure-of-8 microphone is duplicated into two separate tracks where one of them is hard-panned to the left, and the other hard-panned to the right. One of the Side channels is then phase-flipped so they cancel out the opposite side.

 

[Axo1989]

I'm not sure what you two are discussing, but I just want to point out that when it comes to mixing of a Mid/Side recording, it's no longer just one "Side" channel, it's actually two of them. So in Mid/Side mixing, there are three separate tracks in total, one Mid channel and two Side channels.

The recording of the "Sides" made with the Figure-of-8 microphone is split into two separate tracks where one of them is hard-panned to the left, and the other hard-panned to the right. One of the Side channels is then phase-flipped so they cancel out the opposite side.

Not exactly. The sides presented at the mixing console are as I described: M+S=L and M+(-S)=R (panned to L and R respectively). The third channel (depending on your technique) is both of these summed and centred, or just M straight up. It's probably why @LevityProject thinks the side channel is stereo. And yes, it's a tangential discussion vs the thread topic. I just couldn't help myself.

 

[goat76]

Not exactly. The sides presented at the mixing console are as I described: M+S=L and M+(-S)=R (panned to L and R respectively). The third channel (depending on your technique) is both of these summed and centred, or just M straight up. It's probably why @LevityProject thinks the side channel is stereo. And yes, it's a tangential discussion vs the thread topic. I just couldn't help myself.

Maybe you missed that I changed the word “split” to “duplicated” in my previous reply, as it was a typo. Other than that, I think you describe the same thing I do as most of the mathematical details in your description is automatically taken care of by the pan pot in the mixing console or the DAW. It’s just the “summed” part in your description I don't quite understand what you mean, as they are still three separate tracks that can be individually adjusted.

 

[LevityProject]

Not exactly. The sides presented at the mixing console are as I described: M+S=L and M+(-S)=R (panned to L and R respectively). The third channel (depending on your technique) is both of these summed and centred, or just M straight up. It's probably why @LevityProject thinks the side channel is stereo. And yes, it's a tangential discussion vs the thread topic. I just couldn't help myself.

I should probably be more careful ... stereo information is in the side channel. Think about the side channel as stereo (even though it is technically a mono channel that contains stereo information) is incredibly helpful as a frame.

When changing the side channel, it's going to change the stereo information and not the mono information. That's a hard fact.
However, it's also possible to change the perception of stereo information by changing the mid channel - by cutting rather than boosting - either through overall gain reduction, or cutting with an EQ in the mid-channel. This can carve space for the stereo information of the side channel. Anyhow, that's sort of another topic, but illustrates the point that Mid/side is a real relationship between the Left and Right channels.

And it exists in both the digital and analog domains. So it's helpful to think about when judging something like sound stage. Eg How can you raise the perception of an increased side channel on a set of speakers? Either through DSP widening or by pulling the speakers further apart.

By moving speakers further apart, you are effectively changing the relationship of direct sound and indirect sound arriving at your ears. Wider apart, it will take longer for the sound to get to your ears, which leaves more chance for room reflections to interfere (this is not necessarily a bad thing, but can become too overbearing if there isn't enough direct sound). Again, this is just a way of thinking about what is happening and is not an attempt to explain all of the hard science behind it.

The point is, a good speaker setup optimizes the stereo field. Put another way, a good speaker setup optimizes the LR and Mid/Side relationships.

 

[Axo1989]

Maybe you missed that I changed the word “split” to “duplicated” in my previous reply, as it was a typo. Other than that, I think you describe the same thing I do as most of the mathematical details in your description is automatically taken care of by the pan pot in the mixing console or the DAW. It’s just the “summed” part in your description I don't quite understand what you mean, as they are still three separate tracks that can be individually adjusted.

I did miss that. Yes I think we are describing the same thing and I agree the mixer/DAW setup presents channels to manipulate without illustrating the matrix. As for “summed” is “combined” better? The S0S article illustrates:

Spoiler: M-S mixer

The mixer "mid" channel is provided as a means to adjust relative levels of mid vs side to effect stereo image width and monitor the result, so we listen to M-S matrixed to L-R (obviously we can't simply listen to M-S mapped directly to L-R and have it make sense to our ears as a stereo image). I didn't mean to go overboard on this tangent, for some reason I wanted to de-convolve @LevityProject's description for my own peace of mind.

 

[Axo1989]

When changing the side channel, it's going to change the stereo information and not the mono information. That's a hard fact.
However, it's also possible to change the perception of stereo information by changing the mid channel - by cutting rather than boosting - either through overall gain reduction, or cutting with an EQ in the mid-channel. This can carve space for the stereo information of the side channel. Anyhow, that's sort of another topic, but illustrates the point that Mid/side is a real relationship between the Left and Right channels.

I can see that's how you interpret it, but we obviously assign different meanings to phrases like "hard fact". Your understanding is too convoluted for me, when the actual goings-on are more straightforward.

 

[LevityProject]

I can see that's how you interpret it, but we obviously assign different meanings to phrases like "hard fact". Your understanding is too convoluted for me, when the actual goings-on are more straightforward.

That's fair. I really should get my terminology ironed out anyhow. That'll be less confusing

But it's really quite a blessing to be part of a community now where the vast majority of people seem to have a much better technical understadning than I do. I have experiential understanding, but I lack the technical side ... kind of like a musician who can understand the feel and flow of music, but doesn't know music theory.

So, moving ahead I'll just reference mid-side as the mid-channel and side-channel - at least when referencing processing.

And yep I was off on the sides being the "thing" that creates stereo. It must have something to do with phase between L and R that creates the stereo field since adjusting the mid or side (whether digitally or non-digitally) affects phase and either exaggerates or narrows the stereo field. I know it's possible to exaggerate the stereo field so much that the entire signal goes out of phase, which is why I'm coming to that idea.

 

[Emlin]

That's fair. I really should get my terminology ironed out anyhow. That'll be less confusing

But it's really quite a blessing to be part of a community now where the vast majority of people seem to have a much better technical understadning than I do. I have experiential understanding, but I lack the technical side ... kind of like a musician who can understand the feel and flow of music, but doesn't know music theory.

So, moving ahead I'll just reference mid-side as the mid-channel and side-channel - at least when referencing processing.

And yep I was off on the sides being the "thing" that creates stereo. It must have something to do with phase between L and R that creates the stereo field since adjusting the mid or side (whether digitally or non-digitally) affects phase and either exaggerates or narrows the stereo field. I know it's possible to exaggerate the stereo field so much that the entire signal goes out of phase, which is why I'm coming to that idea.

Jeez.

 

[LevityProject]

Jeez.

That isn't fair.

 

[PristineSound]

Here is a snippet from Toole's Sound Reproduction:
". . .spatial quality and sound quality ratings were obviously not completely independent-one followed the trends of the other. Is it possible that listeners cannot separate them even though, consciously, most were confident that they could (the author included [meaning Toole, himself])?"

I am an absolute sucker for spatial effects and characteristics on two channel loudspeaker systems.

But I think grouping "a clear perception of where within this space different sounds are located" with soundstage may muddy the water a bit. I considered imaging, depth localization, holographic effects to be separate categories from sound stage.

As for point source, I do not think point source is absolutely necessary, but I do think it helps tremendously. I wonder if the reason is because point source eliminates lobing, hence better controlled directivity.

Toe in and placing it far from the front wall is all related to placement, and I agree with both, they definitely help with my space.

Room treatment is a must to tame reverberations as it can smear the sound.

Of course controlled directivity is needed otherwise dissimilar sound can also smear the spatial effects.

I also find that a pair match speaker does wonders, they will take spatial effects to the next level.

Lastly, DIRAC is huge for my space in creating a mind bending phantom center.



EDIT: this was moved from a thread that I started without knowledge this thread exist. And I made some edits to my post here.

 

[RayDunzl]

I like the in-your-face gotta sit in the sweet spot of my electrostatic hybrids with an overwhelming amount of power available.

They don't have much in the way of side/floor/ceiling bounces to modify the direct sound.

But for daily use the little JBL LSR 308 fill the economical operation niche for casual listening from TV and PC and HDRadio and whatever.

They spray the room, as evidenced by impulse and unsmoothed frequency response.

 

[sergeauckland]

I tried to post this in the previous thread, but it got moved here.

One characteristic that isn't assessed on ASR is that of pair-matching between loudspeakers of a stereo pair. From what I have measured and understood, the main contributor to accurate stereo imaging, spacial effects etc. is accurate pair matching. As far as I know, only KEF quoted a figure for pair-matching on their Reference Series, although others may also have done so.

Whilst I completely accept Olive and Toole's suggestion that loudspeakers are best evaluated singly, stereo imaging capabilities seem almost totally dependent on accurate pair matching. If one thinks about how a phantom image is formed, a significant difference between left and right varying with frequencies will spread out the stereo image as the music changes frequencies, especially so as most music is made up of many different frequencies at the same time.

Consequently, A loudspeaker without a pair-matching specification is suspect as a stereo (or multichannel) reproducer, however well it reproduces a single channel signal.

Listen to a two channel mono signal. Do you get a tight central image, or it is diffuse somewhere around the centre? That to me is the subjective test of good pair-matching. Of course Kippel measurements are tedious to do, so manufacturers and those few reviewers with such a system will measure just one loudspeaker. That will characterise the loudspeaker as a reproducer, but not as a reproducer of stereo.

S.

 

[Mayan]

Yes, psychoacoustics are very important. For example, listening to music in the dark, with the lights off, will give a very different soundstage compared to listening to music with the lights on.

Wholeheartedly agree with, and when this particular ambient lighting listening session or schedule truly grabs hold of their is absolutely no turning back. Not to mention many of us would agree during late evening hours our electrical grid is less taxed, compromised, and allowing the system a cleaner foundation slightly lowering the noise floor. A quiet room without any distractions disrupting your attention and a silenced backdrop are now in motion.

 

[Salt]

I tried to post this in the previous thread, but it got moved here.

One characteristic that isn't assessed on ASR is that of pair-matching between loudspeakers of a stereo pair. From what I have measured and understood, the main contributor to accurate stereo imaging, spacial effects etc. is accurate pair matching. As far as I know, only KEF quoted a figure for pair-matching on their Reference Series, although others may also have done so.

Whilst I completely accept Olive and Toole's suggestion that loudspeakers are best evaluated singly, stereo imaging capabilities seem almost totally dependent on accurate pair matching. If one thinks about how a phantom image is formed, a significant difference between left and right varying with frequencies will spread out the stereo image as the music changes frequencies, especially so as most music is made up of many different frequencies at the same time.

Consequently, A loudspeaker without a pair-matching specification is suspect as a stereo (or multichannel) reproducer, however well it reproduces a single channel signal.

Listen to a two channel mono signal. Do you get a tight central image, or it is diffuse somewhere around the centre? That to me is the subjective test of good pair-matching. Of course Kippel measurements are tedious to do, so manufacturers and those few reviewers with such a system will measure just one loudspeaker. That will characterise the loudspeaker as a reproducer, but not as a reproducer of stereo.

S.

I would extend to pair matching at listening position, with the room involved, not only the speaker at it's own.

 

[sergeauckland]

I would extend to pair matching at listening position, with the room involved, not only the speaker at it's own.

Ideally yes, but when choosing a loudspeaker to purchase, the only accurate information we may have is pair matching under anechoic conditions. If the loudspeaker isn't accurate under those conditions, then there's little hope for it being any better at home. That's why I consider the anechoic frequency response and pair matching to be an essential specification point. If a manufacturer can't quote their spec for pair matching, then they can't be serious.

S.

 

[Salt]

Ideally yes, but when choosing a loudspeaker to purchase, the only accurate information we may have is pair matching under anechoic conditions. If the loudspeaker isn't accurate under those conditions, then there's little hope for it being any better at home. That's why I consider the anechoic frequency response and pair matching to be an essential specification point. If a manufacturer can't quote their spec for pair matching, then they can't be serious.

S.

Agree: pair matching speaker in a pair matching room / environment and we're clear for spatial listening.

 

[Arindal]

I wonder if the reason is because point source eliminates lobing, hence better controlled directivity.

Most of coaxial point source designs on the market actually have problems with controlling the directivity as they have a tendency to increasing direcitivity index or even show a step up in d.i.

But in general I would agree, more even directivity helps imaging. With point source designs, I would put the fact that there are usually no pronounced cancellation effects at x-over point as of greater importance. Same is the simple geometry, as there are hints that our ears can determine the location of a real sound source by the early reflections originating from it. So particularly designs with midrange and tweeter far apart in close proximity to the listener, might make it more difficult to ensure stable and sharp localization.

when choosing a loudspeaker to purchase, the only accurate information we may have is pair matching under anechoic conditions.

The directivity index over frequency and the polar plots, particular under angles prone to cause early reflections, should also be taken into account. As well as the driver geometry, as mentioned previously.

While I have experienced several factors which make it more likely to have good localization, it is not a guarantee. Particularly if you take depth-of-field, ambience and ´spatial aspects around the phantom sources´ into account, you cannot predict how it will sound, you have to listen to it. Preferably with recordings containing a lot of meaningful reverb pattern, such as sacred vocal music.

Listen to a two channel mono signal. Do you get a tight central image, or it is diffuse somewhere around the centre? That to me is the subjective test of good pair-matching.

That is a very important test, can recommend it. But with compromised center image localization, the outcome of poor pair matching is not the only explanation. In most cases, early reflections causing mirrored phantom sources, or cancellation effects or problems with the real sources becoming localizable, is more likely to cause poor center imaging.

A simple test can differentiate what is the most likely cause: If you listen to the mono center signal noticing poor imaging stability, and then step by step reduce the listening distance, with poor pair-matching as the root you will experience worse localization stability. In contrary, if early reflections, interference effects or angle errors are the main cause, reducing the listening distance will make the center localization more stable while increased listening distance will compromise the staging further (which is not the case with pair differences in FR).