A Broad Discussion of Speakers with Major Audio Luminaries

[levimax]

My setup is capable of high SPL with low distortion and is time-aligned with linear phase crossovers. LFs are not summed to mono and spatial effects are audible down to circa 40Hz. There are certainly even better systems out there, but mine is no slouch. Upmixing is not a kludge to mitigate limitations of the loudspeakers; it is a kludge to mitigate limitations of standard 2-channel stereo.

It is an interesting question why stereo upmixing is often preferred. Is it because it covers up limitations of the loudspeakers / room or that it covers up limitation of stereo itself, or something else? Is there any research that answers that question?

 

[goat76]

It is an interesting question why stereo upmixing is often preferred. Is it because it covers up limitations of the loudspeakers / room or that it covers up limitation of stereo itself, or something else? Is there any research that answers that question?

But is upmixing really “often” preferred, I mean, most people don’t even have a surround system. I would also like to see a research on this.

 

[Satanica]

My setup is capable of high SPL with low distortion and is time-aligned with linear phase crossovers. LFs are not summed to mono and spatial effects are audible down to circa 40Hz. There are certainly even better systems out there, but mine is no slouch. Upmixing is not a kludge to mitigate limitations of the loudspeakers; it is a kludge to mitigate limitations of standard 2-channel stereo.

I think it's a "kludge" if.

The up-mixing algorithm is poor or poorly configured.
The up-mixed speakers don't match or exceed the technical ability of the left and right speakers.
The up-mixed speakers are not level aligned and time aligned to the left and right speakers.
The up-mixed surround speakers are not well positioned e.g. not the same or similar vertical plane, the surrounds are not at 90 degrees or slightly behind the ears.
The up-mixed speakers are not a similar distance to the ears as the left and right speakers.

Also the algorithm really needs to be considered.
Some up mixers especially those in the pro space that can be used at home such as NUGEN Halo Upmix have a lot of options, almost infinite tweakability (dog chasing it's own tail?).

There are many options, considerations, and possible error points when it comes to up-mixing.
And, in general, does it add to the "circle of confusion" feedback loop?

 

[Floyd Toole]

It is an interesting question why stereo upmixing is often preferred. Is it because it covers up limitations of the loudspeakers / room or that it covers up limitation of stereo itself, or something else? Is there any research that answers that question?

I would put it differently. Multichannel audio and good stereo upmixing don't cover up anything. They reveal perceptual dimensions that are deficient in stereo.

If you ever have the opportunity to compare a good multichannel classical music recording with the stereo equivalent or downmix I don't think you would be asking for research to justify multichannel. But, if you want science, there is abundant data relating to the kinds of sounds needed to be delivered to a listener's ears that result in spatial localization of specific sound sources, and of those that result in the very desirable perception of envelopment - of being in a large space, like a concert hall. Some of this is described in my books and papers cited therein. The inevitable conclusion is that stereo is incapable of delivering the necessary sounds at the right times from the right directions. It gets the front part of the 3D live experience quite well but falls short on the ability to transport the listener to another space. We humans are very adaptable, and can make amazing compensations for many things, so we end up satisfied with compromises. The music itself is the truly interesting part of the experience and it is very durable.

 

[levimax]

But, if you want science, there is abundant data relating to the kinds of sounds needed to be delivered to a listener's ears that result in spatial localization of specific sound sources, and of those that result in the very desirable perception of envelopment - of being in a large space, like a concert hall. Some of this is described in my books and papers cited therein. The inevitable conclusion is that stereo is incapable of delivering the necessary sounds at the right times from the right directions

The science I was interested in is preference testing that compares a Stereo signal played in 2 channel to a upsampled version of that same stereo signal. I understand about true multichannel recordings compared to stereo recordings.

 

[suttondesign]

I would put it differently. Multichannel audio and good stereo upmixing doesn't cover up anything. It reveals perceptual dimensions that are deficient in stereo.

If you ever have the opportunity to compare a good multichannel classical music recording with the stereo equivalent or downmix I don't think you would be asking for research to justify multichannel. But, if you want science, there is abundant data relating to the kinds of sounds needed to be delivered to a listener's ears that result in spatial localization of specific sound sources, and of those that result in the very desirable perception of envelopment - of being in a large space, like a concert hall. Some of this is described in my books and papers cited therein. The inevitable conclusion is that stereo is incapable of delivering the necessary sounds at the right times from the right directions. It gets the front part of the 3D live experience quite well but falls short on the ability to transport the listener to another space. We humans are very adaptable, and can make amazing compensations for many things, so we end up satisfied with compromises. The music itself is the truly interesting part of the experience and it is very durable.

this. i compare it to watching baseball on tv vs live. both are awesome.

 

[bmc0]

I think it's a "kludge" if [...]

Perhaps the wrong choice of word on my part. I upmix nearly all of the two-channel recordings I listen to. What I mean to say is that discrete multichannel is technically a better solution if available. Most of your points are not specific to upmixing and apply to discrete multichannel as well.

 

[Satanica]

I pretty much always prefer multi-channel mixes over stereo of any genre, even with (my) what objectively would be called a compromised surround setup.
But it gets a bit more complicated, many multiple-channel productions are more dynamic than the equivalent stereo mix that have been subjected to LOUDNESS WAR compression. So, advantage multi-channel straight up I notice most of the time, especially with rock/pop/metal/hip-hop/electronic "modern" genres.

For non concert hall studio recordings, there is much less recorded space which is typical within modern (see above) and most popular genres.

Overall, I perceive the vast majority of artists don't care about acoustic space on their recordings, they want the listener to concentrate on the actual music and words and meaning, rather than the "acoustic canvas" that their art was audibly "performed on". Granted, for genres such as classical and jazz, acoustic space is seemingly much more of a thing both for the artists and recording engineers.

 

[Satanica]

Perhaps the wrong choice of word on my part. I upmix nearly all of the two-channel recordings I listen to. What I mean to say is that discrete multichannel is technically a better solution if available. Most of your points are not specific to upmixing and apply to discrete multichannel as well.

Yep agreed of course. My overall point is, adding on extra channels for up-mixing (or discrete) needs to be taken seriously to get a good or even better result than just two speakers, the extra channels can't be just slapped on add-ons and then everything instantly becomes better.

 

[Newman]

[to Floyd Toole] The science I was interested in is preference testing that compares a Stereo signal played in 2 channel to a upsampled version of that same stereo signal. I understand about true multichannel recordings compared to stereo recordings.

He did say, "…and good stereo upmixing".

 

[Floyd Toole]

The science I was interested in is preference testing that compares a Stereo signal played in 2 channel to a upsampled version of that same stereo signal. I understand about true multichannel recordings compared to stereo recordings.

OK, I understand the objective, but elsewhere in these forums I have said that upmixing is a lottery because of the numerous uncontrollable variables: the recording, the upmixer, the directivity patterns of the loudspeakers and geometry and reflectivity of the rooms, and the personal preferences of the judges/listeners. It is a fundamentally confused scenario in which the outcome is not assured. No scientific research is required to prove it.

That is why in my earlier post I said "good stereo upmixing" because there are so many opportunities for bad stereo upmixing. In my experiences I have had to manipulate the parameters of the upmixer (when it was possible) to optimize results for individual recordings. Then there were recordings for which the only satisfying setting was "off". This is clearly a topic area for experimenters who have patience.

At one time, not many years ago, I would have argued that the audio industry should get its act together and create a standardized system that has some assurance of having the art "translate" to homes in all of its aspects. That quaintly old school perspective assumed that significant numbers of homes had recognizable stereo systems, such as appears to be the case with ASR forum members. But we are vastly outnumbered by people who are adequately rewarded by listening through soundbars, smart speakers, headphones and in cars.

The common element is the music, which "translates" sufficiently through all these delivery systems to provide enjoyment. From the sound quality perspective, science and technology have resulted in much improved and sometimes quite impressive results at modest cost. From the "soundstage and imaging" perspective, there is great variability. So, if there is an interesting line of enquiry for science at the moment, it would be to ascertain just how important those elements of perception are in the total satisfaction listeners derive from listening to music.

In my present living situation I actually spend some of my time with my laptop in my lap, my feet up, seated off axis, where I can enjoy a nice view through our picture windows, listening to high quality stereo reproduced in a room with no bothersome resonances. Ideal? Not at all. Enjoyable? The music gets through very well. The sound quality is good. There is a sense of ambiguous "space". I know what the missing spatial elements should sound like, and my imagination fills in the blanks. I adapt to the circumstances.

Soundstage and imaging are not part of the music, they are part of a narrow, very traditional, set of expectations based on live performances experienced from the expensive seats, near center audience and not too far from the orchestra. In typical concert halls only a small fraction of the audience gets to hear it. Where one sits makes an enormous difference, and few can afford the good seats. Stereo, even mediocre stereo, may be better than what is heard by many live audience members in affordable seats. Fortunately the sense of large space envelopment is more generously perceived throughout the audience, but not everywhere in all halls. Attending a live classical concert is no assurance of hearing all that there is to hear.

Both stereo and multichannel sound had their origins as enhancements to movies, and for me at least, movies benefit significantly from multichannel reproduction. Putting the viewer into the acoustical space that is portrayed on the screen is an obvious enhancement. Again, only a fraction of the cinema audience gets to hear the full splendor of the soundtracks, so a good home theater may be an improvement (to me it has been in both sound and spatial qualities). I often wonder why so many people gravitate to the side or rear seats in cinemas - so much is missed. Stereo and headphone reproduction provide useful spatial/directional cues, but are compromised. Mono succeeds only in conveying dialogue and plot. Movie mixers have worked with multichannel audio for decades and have become good at it. Music mixers schooled in stereo have demonstrated some problems adapting, but a new generation is coming along.

So, perhaps stereo upmixing cannot deliver predictably beneficial outcomes from all recordings in all playback situations, but it may still be a worthwhile feature to have available.

 

[Duke]

From the sound quality perspective, science and technology have resulted in much improved and sometimes quite impressive results at modest cost. From the "soundstage and imaging" perspective, there is great variability. So, if there is an interesting line of inquiry for science at the moment, it would be to ascertain just how important those elements of perception are in the total satisfaction listeners derive from listening to music.

My recollection is that in your book, or at least in the 3rd edition thereof, you cite a study by Wolfgang Klippel that might be relevant. His findings included an assessment of the subjective role of the "feeling of space", rather than "soundstage and imaging", and those may not be exactly the same thing, but both have to do with spatial quality rather than sound quality.

Klippel found that “naturalness” (realism and accuracy) was 30% related to sound quality (coloration, or the lack thereof); 20% related to tonal balance; and 50% related to the “feeling of space”.

And “pleasantness” (general satisfaction or preference) was 30% related to sound quality and 70% related to the “feeling of space”.

So the "feeling of space" plays a big (50%) role in "naturalness" and an even bigger (70%) role in "pleasantness", by his definitions of those words. My guess is that the role "soundstage and imaging" plays would probably be along similar lines, but I don't know, as the exact definitions might exclude some aspects of "the feeling of space".

And... there may even be a direct link between perceived accuracy and imaging! Francis Rumsey (citing Gunther Theile) describes what he calls "decoloration", a phenomenon whereby timbral coloration (or at least coloration due to stereo's comb filtering) decreases as the ability to detect spatial images improves.

Anyway my point is, I think there is already evidence that spatial quality matters a lot, and can even play a direct role in sound quality. I don't think we necessarily need to wait for another study to answer the question of whether, and approximately how much, spatial quality (or "the feeling of space") matters.

 

[Floyd Toole]

My recollection is that in your book, or at least in the 3rd edition thereof, you cite a study by Wolfgang Klippel that might be relevant. His findings included an assessment of the subjective role of the "feeling of space", rather than "soundstage and imaging", and those may not be exactly the same thing, but both have to do with spatial quality rather than sound quality.

Klippel found that “naturalness” (realism and accuracy) was 30% related to sound quality (coloration, or the lack thereof); 20% related to tonal balance; and 50% related to the “feeling of space”.

And “pleasantness” (general satisfaction or preference) was 30% related to sound quality and 70% related to the “feeling of space”.

So the "feeling of space" plays a big (50%) role in "naturalness" and an even bigger (70%) role in "pleasantness", by his definitions of those words. My guess is that the role "soundstage and imaging" plays would probably be along similar lines, but I don't know, as the exact definitions might exclude some aspects of "the feeling of space".

And... there may even be a direct link between perceived accuracy and imaging! Francis Rumsey (citing Gunther Theile) describes what he calls "decoloration", a phenomenon whereby timbral coloration (or at least coloration due to stereo's comb filtering) decreases as the ability to detect spatial images improves.

Anyway my point is, I think there is already evidence that spatial quality matters a lot, and can even play a direct role in sound quality. I don't think we necessarily need to wait for another study to answer the question of whether, and approximately how much, spatial quality (or "the feeling of space") matters.

There is considerable ambiguity in what is meant by "spatial quality" because if it includes all aspects of "space" it includes specific images localized somewhere in space, which are placed in the context of a frontal soundstage (the band/orchestra) both of which are perceived as existing within a larger space that includes the listener, commonly understood to be envelopment. So there is no doubt whatsoever that "spatial quality" matters a great deal and no additional research is necessary. The 4th edition has a chapter on it alone, which segregates the "envelopment" aspect from the frontal soundstage and imaging that dominates stereo, and that multichannel truly liberates.

What I was getting at was the frontal experience itself, that really only exist in live performances for the monied minority - it is an "elitist" criterion, if you like. I'm not arguing against it, because I enjoy it and can afford it, but I suspect that more have experienced it in even flawed stereo reproduction than ever got to hear it in a good concert hall from the best seats. So, how much does it add to our musical enjoyment? Some, for certain, but how much? And why are so many people satisfied with greatly corrupted versions, or even none of it?

 

[Duke]

There is considerable ambiguity in what is meant by "spatial quality" because if it includes all aspects of "space" it includes specific images localized somewhere in space, which are placed in the context of a frontal soundstage (the band/orchestra) both of which are perceived as existing within a larger space that includes the listener, commonly understood to be envelopment. So there is no doubt whatsoever that "spatial quality" matters a great deal and no additional research is necessary. The 4th edition has a chapter on it alone, which segregates the "envelopment" aspect from the frontal soundstage and imaging that dominates stereo, and that multichannel truly liberates.

This makes sense; I was failing to differentiate "envelopment" from "the frontal soundstage". Thank you for your precision and patience!

 

[krabapple]

I would put it differently. Multichannel audio and good stereo upmixing don't cover up anything. They reveal perceptual dimensions that are deficient in stereo.

If you ever have the opportunity to compare a good multichannel classical music recording with the stereo equivalent or downmix I don't think you would be asking for research to justify multichannel. But, if you want science, there is abundant data relating to the kinds of sounds needed to be delivered to a listener's ears that result in spatial localization of specific sound sources, and of those that result in the very desirable perception of envelopment - of being in a large space, like a concert hall. Some of this is described in my books and papers cited therein. The inevitable conclusion is that stereo is incapable of delivering the necessary sounds at the right times from the right directions. It gets the front part of the 3D live experience quite well but falls short on the ability to transport the listener to another space. We humans are very adaptable, and can make amazing compensations for many things, so we end up satisfied with compromises. The music itself is the truly interesting part of the experience and it is very durable.

Also, I seem to recall from your book, that just as a single speaker is more revealing of its flaws that the same speaker in 'stereo', more delivery channels mask what might be noticeable in two.
Certainly my experience is that auditioning a single surround channel of an upmix can reveal artifacts that I simply don't notice in normal listening to that upmix.

 

[Floyd Toole]

Also, I seem to recall from your book, that just as a single speaker is more revealing of its flaws that the same speaker in 'stereo', more delivery channels mask what might be noticeable in two.
Certainly my experience is that auditioning a single surround channel of an upmix can reveal artifacts that I simply don't notice in normal listening to that upmix.

Yes, you are right, but it needs to be made clear that adding channels, from mono to stereo to multichannel the aspect that is "masked" or more accurately is more difficult to hear, is loudspeaker resonances, but not for hard panned components within the mixes. Otherwise, adding channels adds the ability to present images in different locations and the potential for gratifying envelopment.

 

[goat76]

Anyway my point is, I think there is already evidence that spatial quality matters a lot, and can even play a direct role in sound quality. I don't think we necessarily need to wait for another study to answer the question of whether, and approximately how much, spatial quality (or "the feeling of space") matters.

I don't doubt for a second that spatial quality matters to many listeners.

What I doubt is that upmixing 2-channel mixes is widely preferred by most listeners, since the result depends a lot on how the 2-channel mix was done. Some mixes are done to try to recreate/document a full musical event as if the listener were there in person, while others focus more on the sound quality of individual sound objects panned for effects or maximal separation. One single "static" upmixing algorithm can simply not sort them out equally well, making the upmixing either a hit or a miss depending on the material.

The music I listen to most is usually semi-realistically made, meaning there are some things in the mix aimed at recreating a three-dimensional space, while the mix also contains panned single elements/instruments within that space. I don't find upmixing works very well for that type of mix.

 

[Mort]

Works great for electro -pop

 

[audiofooled]

Stereo can have considerable spatial quality but it requires manipulation of indirect sound, including very clean direct path.

ETC where both early and late reflections are largely inaudible:

The same room, with audible late reflections (and some early ones):

First 10 ms of the first graph:

First 10 ms of the second graph:

Late arrival of the first graph:

Late arrival of the second graph:

Distance between front and rear wall:

I thought it may be interesting to someone because it's the same room and the same main listening position, but different sound field manipulation.

 

[krabapple]

I don't doubt for a second that spatial quality matters to many listeners.

What I doubt is that upmixing 2-channel mixes is widely preferred by most listeners, since the result depends a lot on how the 2-channel mix was done. Some mixes are done to try to recreate/document a full musical event as if the listener were there in person, while others focus more on the sound quality of individual sound objects panned for effects or maximal separation. One single "static" upmixing algorithm can simply not sort them out equally well, making the upmixing either a hit or a miss depending on the material.

Playing stuff in stereo is hit or miss too, for the same reasons: recording quality, mix quality, EQ, and of course performances themselves, vary.

The music I listen to most is usually semi-realistically made, meaning there are some things in the mix aimed at recreating a three-dimensional space, while the mix also contains panned single elements/instruments within that space. I don't find upmixing works very well for that type of mix.

I do. I enjoy both the simple expansiveness/envelopment upmixing lends to many 'classical' (acoustic instruments in 'real' spaces) recordings, as well as the similar trick it can perform on studio work, where as a bonus there's the occasional surprising/fun placement of a pan-potted or time-distorted (e.g. phase shifted) element. (I wouldn't dream of listening to Jimi Hendrix these days without upmixing.)

And as I've said before, I not infrequently prefer an upmix of a stereo mix I already like, to someone else's attempt at remixing it in surround from multitracks.