This is exactly how I feel about it! Thanks.
Then we might come to a conclusion?
Do you really think there's nothing more to learn? The more I learn about the subject, the more questions arise. I still don't know how to design 'the ultimate stereo listening experience', the one that can not be surpassed. Some systems that are designed by 'the rules' sound good but not great, but other systems that aren't designed by those rules may sound good too. I'm still not sure what makes a truly great system, though.
I mentioned accuracy in post 1469.
Sharp imaging, I think, is the key issue is this whole directivity discussion. Sharp imaging is required in the production end, as most recordings are done with a microphone placed close to the instrument and positioned in the acoustic space afterwards. It is not the 'original event', but a production.
As we know from this discussion: magnitude, time-delay, direction and number of reflections have influence on imaging and envelopment, exact numbers are unknown. I would say that at the reproduction end the loudspeakers must be able to deliver the same sharpness in imaging as the production end, just like a 1080P HD movie doesn't come to it's full potential on a 20 inch SD television. Omni's, dipoles and conventional boxes are all perfectly capable of delivering this sharp image, but in general they need more distance to the walls to deliver the same sharpness as the narrow directivity waveguide speakers can deliver. So, room size dictates the directivity pattern.
Remains spaciousness and ASW. Both should, in my opinion, be controlled at the prodcution side. It can be done, a nice example is this one:
"Room size dictates directivity", I think there's truth in that. Ultimately the sound of a speaker/room-combination can be reduced to the direct sound and the total reflection pattern. So maybe in the opening post I asked the wrong question. It shouldn't have been about the ideal directivity pattern, but about the ideal reflection pattern.
Floyd Toole supposedly has the answers; they're in his book. Research indicates that early reflections aren't as bad as has long been thought. But those early reflections cause several different effects. It very much depends on what questions you ask the subjects if the their effect will be reported to be beneficial or not. It probably also depends on what qualities you find most important in a hifi system. For me tone coloration and listener fatigue are the worst offenders. I'd gladly give up some ASW or spaciousness for it.
If sound quality still is (partially) a matter of personal taste, then what about Toole's circle of confusion? Can it ever be broken? He advocates using accurate speakers in a good room to hear what's really on the recording, but how can we ever do that if we can't agree on what accurate is?
Yes by standardization.
You can hear what's on the recording by using headphones. Accuracy defined as "hearing everything that's on a recording" is the RFZ, Geddes, anechoic chamber, etc. approach. Is it an accurate representation of the live event? No and it doesn't have to. But if you're looking for "realism", then more "exotic" concepts are appropriate. One of these "esoteric" concepts is multichannel
It is true that one can hear what is on the recording using headphones, but the chance is big that it is not as intended by the producer/sound engineer because the mix is made for 2 loudspeakers and not for headphones. So "hearing everything that's on a recording" should not be the definition of accuray, but "hearing like the producer/engineer heard it at the production stage" should be the definition of accuracy. So if the engineer decides to place the piano at 90 degrees at the right side using some crosstalk-cancellation technique, this should be reproduced at the reproduction side. Nothing more, nothing less.
Is it an accurate representation of the live event? No and it doesn't have to. But if you're looking for "realism", then more "exotic" concepts are appropriate. One of these "esoteric" concepts is multichannel
Agreed!
I agree but it is a hopeless endeavor without extensive standardization. In music production there's not even a standardized reference level and this is more than 80 year after people started recording music...
Let's not forget that Toole points to multichannel surround sound as the solution to the problems stereo has. He points out that in a surround system, differences in the polar pattern of the speaker are probably much less audible or problematic.
I think that if you consider only two speakers, there is a continuum of "good speakers" that have their own strong and weak points (for example, imaging sharpness vs. ASW/spaciousness). I think that this thread has already brought very good insights into the trade-offs and psychoacoustics involved.
I agree the concept surround sound can solve many of the problems we face with stereo, yet still I'd like to find where the limits of stereo are, i.e. what hypothetical conditions would allow reproduction closest to perfection. So, that's what kind of speakers should you use, in what kind of room and how should you set them up.
Then we might come to a conclusion?
Not until I've expressed all of my opinions at least 6 times. Only half way there.
David S.
I think that reproduction in a large, relatively reverberant room with directional CD speakers might be the way to go for pure stereo. It would more or less meet the "less early, more later reflections" idea. Minimize < 10 ms, maximize > 10 ms. You would get sharp imaging, minimal colouration and an early spatial impression from the later reflections (probably they still cannot not really be called "late" reflections, needed for envelopment, which should arrive after > 30 - 80 ms).
However, the delay of the later reflections is the primary issue in a small room. In a very small room you probably have to make a trade-off between sharp imaging and ASW/early spatial impression, since the situation above may be impossible to achieve. The question is then which you prefer, since you may not have them both.
I think we can conclude that one should eliminate the floor bounce in every design solution, since it does not seem to contribute any positive effect.
I would take some issue with the suggestion that a large amount of late reflections are not detrimental to sharp imaging.
I think late reflections are certainly less damaging than early coherent reflections from side-walls and floor to sharp imaging, however if the overall level of reverberant field in the room is too high compared to direct field at the seating location I find that the imaging is not sharp.
It doesn't necessarily "pull" or "stretch" the image in the same obvious way as an early side-wall reflection, but rather "dilutes" its sharpness and can make it much more diffuse.
For example in my room which is relatively small, I sit fairly close to the back wall, and I find that the back room corners, which are slightly behind me (approximately 120 degrees from the forward centre line) play a large role in how well I perceive imaging and how sharp it is.
If I leave both corners bare the imaging is a lot more diffuse and indistinct as I'm relatively far from the speakers and probably not experiencing a very good direct/reverberant ratio. (I really should attempt to measure it)
Speaker directivity doesn't really help here as the rear room corners are not that far off axis from the speakers. If I add a bit of damping in both corners the sharpness of imaging and phantom channel imaging is dramatically better.
I don't know what the threshold figure is in terms of direct/reflected ratio, as I haven't attempted to measure it but there is definitely a threshold effect.
Now you could argue that in my small room the delay from the rear corners to the listener is only about 6ms - whether that counts as an early or a late reflection may be up for debate, (it's at least twice as long as my early side-wall reflections) but I've noticed the exact same effect in larger rooms - a bit of reverberation up to a point is fine, but there is a threshold once crossed where imaging is fairly rapidly diluted as direct/reflected ratio falls, regardless of the time delay.
Another thing I notice is that in a room where you're quite close to this critical direct/reflected ratio, (you're sitting close to the critical distance) that my perception of whether I'm within or beyond the critical distance for the room/speaker set-up can change from day to day.
I'm not sure if its due to small changes in the room acoustics due to leaving a door open vs closed, a blind up vs down, moving items like cushions and chairs around etc, or whether it's purely a psychological/perceptual change, but I have definitely observed days where I feel like I am sitting well into the direct field and imaging is pin-point and strong, and other days where it seems as if I am back into the diffuse/reverberant field beyond the critical distance, with a lot of "room effect" in the sound and less than ideal imaging despite sitting in the same location.
If there is good "safety margin" of direct to reverberant ratio to put the critical distance well beyond my listening distance, I don't notice variations in perception like this.
As well as amplitude and delay of early reflections, I think having a sufficient direct to reverberant field ratio at the listening point is also very important, and this ties into the RT60 as well as the speaker directivity.
Last edited:
I agree, and I don't expect any standard for the next 80 years. But it makes the discussion very difficult.
Perhaps it is necessary to make a classification based on genre (large orchestral, small classical chamber music / jazz, live pop concert, studio pop), because within a genre the recording/production practices are, more or less, similar.
For studio pop, it is the 'wild wild west', but for jazz and large orchestral it could be possible to define an optimized stereo speaker format. For the first category imaging is not important, but envelopment and ASW is whereas for the second one would like to have much sharper imaging.
This would rule out a single type of speaker solution, but perhaps it is the way it is. If you see what is 'on the market' this just might be true, our goal would be to make the smallest subset possible
It was not my intention to suggest that later reflections are not detrimental to imaging (only to a smaller degree indeed). I was trying to define an optimal trade-off between imaging and spaciousness. I assume that an optimal, most realistic solution has both qualities. However tastes can be very different so some may consider even those later reflections unwanted.
For most recordings and most listeners however, it is probably not a bad thing. Whether it makes reproduction more realistic probably also depends on the recording venue.
Of course, multi-channel surround provides the solution if directional front speakers and a less reverberant room are used. In that case, additional late reflections can be added if needed, depending on the recording.
Changing the subject for a moment, whilst playing with speaker positioning tonight something has occurred to me - I think there is an implicit assumption in the conversation that everyone in this thread knows all the tricks about how to "set up" speakers, in terms of positioning, and is equally as good at doing it.
Perhaps this is not the case, and some of us know tricks that others don't, which could be partially responsible for some of the different perceptions of stereo, particularly including difficulty in achieving or perceiving good central phantom imaging at high frequencies.
There aren't that many parameters - angular separation of the speakers from the listener, distance from listener to speakers, toe in, and position and symmetry in relation to walls, but they all interact in complex and sometimes difficult to predict ways, and differently for different speaker designs.
I think we all have a fair idea what toe in and separation do, but my shuffling of speakers tonight reminded me of something that is extremely important, yet difficult to get right without the right testing technique.
And that is ensuring that both speakers are exactly the same distance from the listener relative to the centre line of the triangle.
One speaker being as little as half an inch too far forward or back relative to the other speaker is enough upset the phantom imaging in the upper midrange/treble region, and it's difficult to determine this accurately enough by measuring tape alone, even if there are nearby symmetrical room boundaries to measure to.
On the face of it half an inch or better seems a ridiculous requirement and some might be tempted to say "just sit slightly to one side where they're equidistant and you won't notice the difference". The problem is for whatever reason, this doesn't work, although I'm not entirely sure why.
I think its because if you move to the side to correct the unequal distance you're now off axis from the speakers by differing amounts, causing asymmetry in the response of each speaker, especially if there is significant baffle diffraction causing variations in the near off axis response in the treble.
The point is, if you sit along the centre line when one speaker is too far away from you there will be significant comb filtering - the more the error the lower in frequency it goes and the more noticeable it becomes. Moving to one side sort of helps but not really - the end result is the sweet spot becomes very narrow and temperamental, with discrete "lumps" of sweet spot instead of one contiguous sweet spot. If the distance error is great enough high frequencies can tend to localize to one side or the other rather than in the middle.
Getting the speakers equidistant to the listener within less than half an inch whilst still keeping them symmetrically placed in terms of width and toe in is the key. This is best done by listening and mic measurement, rather than tape measure IMO.
If you play mono pink noise you can locate the sweet spot by listening for comb filtering as you move your head sideways but the problem is if the speaker distance error is more than about an inch its ambiguous as to which one of the "sweet spots" you might encounter is the correct one - there will be several different ones perceived as you move within a space of a couple of feet, but only one of them will truly be free of comb filtering.
If you move the speakers to align with the wrong one you won't get good imaging.
What I've found works well is to use an impulse response measurement to get it as close as possible, then go back to an audible pink noise test to home in on the exact position.
So I set up a microphone on a stand set dead centre at the listening position, turn the balance control on the amplifier partially to one side to give about 6dB attenuation on one channel, and measure the impulse response with both speakers playing at once.
Unless the distance error is very small you'll see two distinct impulse responses, and because one is half the amplitude its easy to tell by comparing their height which speaker is too close or too far. I then adjust the distance of one of the speakers along the listener to speaker axis until the impulse responses perfectly overlap and form one impulse that looks the same as one speaker operating on its own.
That gets us pretty close, but because we listen with two displaced ears I then play mono pink noise and listen carefully for notches in the upper midrange/treble region as I move laterally. Somewhere to each side you'll hear a notch, and in the middle there should be no notch.
If you can see visually that you're lined up symmetrically with both speakers and the pink noise sounds right at the same time you're done.
Despite perhaps having moved the speakers only half to one inch from their initial position determined by a measuring tape, it can often make a huge improvement to the imaging, and make the sweet spot considerably wider.
Does anyone else have any recommended speaker set-up procedures or tips to share ?
As we've seen through the discussion, the psychoacoustics of what is going on in a room and how we perceive it as the speaker set-up is changed is actually pretty complex, and knowing how to get the best out of speakers in a room by setting them up correctly is non-trivial, even for those of us who have a bit of experience with it.
Perhaps this is not the case, and some of us know tricks that others don't, which could be partially responsible for some of the different perceptions of stereo, particularly including difficulty in achieving or perceiving good central phantom imaging at high frequencies.
There aren't that many parameters - angular separation of the speakers from the listener, distance from listener to speakers, toe in, and position and symmetry in relation to walls, but they all interact in complex and sometimes difficult to predict ways, and differently for different speaker designs.
I think we all have a fair idea what toe in and separation do, but my shuffling of speakers tonight reminded me of something that is extremely important, yet difficult to get right without the right testing technique.
And that is ensuring that both speakers are exactly the same distance from the listener relative to the centre line of the triangle.
One speaker being as little as half an inch too far forward or back relative to the other speaker is enough upset the phantom imaging in the upper midrange/treble region, and it's difficult to determine this accurately enough by measuring tape alone, even if there are nearby symmetrical room boundaries to measure to.
On the face of it half an inch or better seems a ridiculous requirement and some might be tempted to say "just sit slightly to one side where they're equidistant and you won't notice the difference". The problem is for whatever reason, this doesn't work, although I'm not entirely sure why.
I think its because if you move to the side to correct the unequal distance you're now off axis from the speakers by differing amounts, causing asymmetry in the response of each speaker, especially if there is significant baffle diffraction causing variations in the near off axis response in the treble.
The point is, if you sit along the centre line when one speaker is too far away from you there will be significant comb filtering - the more the error the lower in frequency it goes and the more noticeable it becomes. Moving to one side sort of helps but not really - the end result is the sweet spot becomes very narrow and temperamental, with discrete "lumps" of sweet spot instead of one contiguous sweet spot. If the distance error is great enough high frequencies can tend to localize to one side or the other rather than in the middle.
Getting the speakers equidistant to the listener within less than half an inch whilst still keeping them symmetrically placed in terms of width and toe in is the key. This is best done by listening and mic measurement, rather than tape measure IMO.
If you play mono pink noise you can locate the sweet spot by listening for comb filtering as you move your head sideways but the problem is if the speaker distance error is more than about an inch its ambiguous as to which one of the "sweet spots" you might encounter is the correct one - there will be several different ones perceived as you move within a space of a couple of feet, but only one of them will truly be free of comb filtering.
If you move the speakers to align with the wrong one you won't get good imaging.
What I've found works well is to use an impulse response measurement to get it as close as possible, then go back to an audible pink noise test to home in on the exact position.
So I set up a microphone on a stand set dead centre at the listening position, turn the balance control on the amplifier partially to one side to give about 6dB attenuation on one channel, and measure the impulse response with both speakers playing at once.
Unless the distance error is very small you'll see two distinct impulse responses, and because one is half the amplitude its easy to tell by comparing their height which speaker is too close or too far. I then adjust the distance of one of the speakers along the listener to speaker axis until the impulse responses perfectly overlap and form one impulse that looks the same as one speaker operating on its own.
That gets us pretty close, but because we listen with two displaced ears I then play mono pink noise and listen carefully for notches in the upper midrange/treble region as I move laterally. Somewhere to each side you'll hear a notch, and in the middle there should be no notch.
If you can see visually that you're lined up symmetrically with both speakers and the pink noise sounds right at the same time you're done.
Despite perhaps having moved the speakers only half to one inch from their initial position determined by a measuring tape, it can often make a huge improvement to the imaging, and make the sweet spot considerably wider.
Does anyone else have any recommended speaker set-up procedures or tips to share ?
As we've seen through the discussion, the psychoacoustics of what is going on in a room and how we perceive it as the speaker set-up is changed is actually pretty complex, and knowing how to get the best out of speakers in a room by setting them up correctly is non-trivial, even for those of us who have a bit of experience with it.
Last edited:
I had the opportunity to listen to the Orions at BAF last weekend. They were sharing the room with a pair of hedlund horns and were being moved in and out from the main spot. Suffice to say, the Orions did not get repositioned exactly where they were last. This, however, did not stop them from creating a very believable image with extreme depth and good separation of instruments. This at various listening positions within the room and with people sitting in front of me. Sure, the imaging suffered a little when you are off the sweet spot, but 99% of it was still there, regardless of whether they were within a inch of each other or if I was sitting right in the middle.
So, no, I do not believe it is so crucial to set up speakers with such meticulousness.
What was perhaps aiding the Orions was that they were about 5 feet from the sidewalls and easily over 6 feet from the rear. This, IMO, is more important to imaging that sitting in the middle or having a perfect equilateral triangle with the speakers. It was kind of a reverberant room, not the best for audio. But they still did a fantastic job. What they did best was disappear. You are only left with the music.
So, no, I do not believe it is so crucial to set up speakers with such meticulousness.
What was perhaps aiding the Orions was that they were about 5 feet from the sidewalls and easily over 6 feet from the rear. This, IMO, is more important to imaging that sitting in the middle or having a perfect equilateral triangle with the speakers. It was kind of a reverberant room, not the best for audio. But they still did a fantastic job. What they did best was disappear. You are only left with the music.
If you want to exactly set the speaker to listener distance and have pink noise and an RTA, the easier approach is mono pink noise out of phase. It will cancel at the microphone and allow very exact setting. If the distances are somewhat close then the low frequencies will generally cancel and as you adjust the position of one or the other speakers you can get it to cancel up to the highest frequencies. You can also do this by ear, but it is more precise to turn sideways and block your far ear. Swing forwards and backwards slightly (perpendicular to the speakers or parallel to the couch). Let someone mark the spot (stand with a finger in the air at the position) then you can turn and see if you are centered on that spot.
As I type this I'm thinking: "Don't do this in public or they'll have you committed."
The sideways technique works just as well if a helper is sliding one speaker forewards and backwards.
Sadly, you will realize that the noticeable accuracy is such that both ears will be significantly off the null point.
While you are measuring with out of phase noise, take a look at how low the carefully cancelled spectrum is. In an anechoic chamber we could allign a pair of speakers. If we start with one speaker and add the second (pink noise in phase) then the curve rises 6dB. Flip the polarity of one speaker and see total cancelation. The broadband curve drops to nothing.
Now in our living room we get perfect cancellation of the direct sound but not of the random diffuse field. The out of phase (but time alligned) cancelation curve is a direct measure of the level of reverberent field. It is set by room reverberence, speaker directivity and listening distance. It is at the crux of whether we sense that our system is very revealing but rather dry, nicely spacious but rather vague, or somewhere in between.
Now I want you all "to get up, get out of your chair and go to the window", no, go to the spectrum analyzer and try the in-phase, out-of-phase difference test and we will see who has the "wettest" and "driest" rooms.
David
I agree, and I don't expect any standard for the next 80 years. But it makes the discussion very difficult.
Perhaps it is necessary to make a classification based on genre (large orchestral, small classical chamber music / jazz, live pop concert, studio pop), because within a genre the recording/production practices are, more or less, similar.
For studio pop, it is the 'wild wild west', but for jazz and large orchestral it could be possible to define an optimized stereo speaker format. For the first category imaging is not important, but envelopment and ASW is whereas for the second one would like to have much sharper imaging.
This would rule out a single type of speaker solution, but perhaps it is the way it is. If you see what is 'on the market' this just might be true, our goal would be to make the smallest subset possible
There's basically two different techniques, one is pan pot recordings (relying on interchannel level differences), the other is AB stereo (relying on interchannel time differences) recordings (and anything in between). You'd need to categorize single recordings and not genres. I'd say that pop is more uniform than the rest (at least within certain decades).
Last edited:
I hadn't heard of this other link until now. The way I see it, we've got a bunch of mostly very adept audio engineers at a table here, with a subject that has been talked mostly to death. Perhaps the best reason to split the multichannel talk off to another thread is so newcomers can find it in the table of contents. I hadn't been able to find any catagory or thread on surround channel extraction, so when people brought it up in this thread, I jumped on it.guys we are drifting off topic again.
I'm not sure whether we posted the link to the other thread we created when we first split off the more esoteric stuff. http://www.diyaudio.com/forums/mult...ques-achieving-oustanding-stereo-imaging.html
Tony.
I've always had best results by toeing the speakers such that they are aimed at a point a foot or two behind my main listening spot. Theoretically, I should want to aim them directly at my head, so as to minimize the relative amplitude of room reflections, which are more than likely detrimental, but I find that the stereo imaging effect is significantly better if I aim the speakers just a few degrees further out. They're still toed in, but just not all the way to my head location. This general rule of thumb has worked in most room I've lived in, but may not work well when the side walls are close in. In a very small room or a car, I would guess that all drivers would be best off aimed directly at your head, to minimize those early (<6mS) reflections.
In my apartment, I always prefer to have my speakers on the long wall of the room if possible. You could talk theory on this for days (and we probably will), but this is the nutshell version of my experience.
Is there even any good sounding live pop concert that would be worth recording? I've seen Dave Matthews in NY lately. After years of not attending any bigger live concert I thought that there would be any advances in sound quality. Unfortunately it's still the same crappy sound. Nowhere near the artistic depth a studio recording can provide. So I'm always wondering why some people want their stereo to sound like a live concert. Do they even listen with their ears or just with their eyes? At DMB they probably just smoke enough
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.