It should read "Green is the direct sound. Blue is the first side wall reflection for the left speaker, red for the right speaker."
I often see figures tossed around like Linkwitz's 6ms to first reflection, or Earl's 10ms to first reflection, but they frustrate me no end because they seem completely unrealistic.
Lets get real here - how on earth can these figures be achieved in anything other than a huge (by European standards) living room ? Sure, maybe some of you in North America have enormous living rooms that give you freedom to put speakers a long way from the walls, but what about the rest of the world ?
Just looking at my current room, 3.45m wide x 4.8m long. Listening distance is dictated to a fixed 2.5 metres, and due to other room obstructions I can't even get the speakers the same distance from the side-walls. Lets work out some numbers.
The right speaker is 0.7m from side-wall to centre axis of the speaker - yes, much closer than I would like but I have no choice. At the listening position this gives me a reflection arriving at 2.74ms, far short of 6-10ms.
The left speaker is 0.9m from side-wall to centre axis, giving 3.69ms. Better but still well short of the target.
What about the floor bounce ? The mid-bass driver is currently 75cm above the floor, that means the floor bounce is delayed by 1.5ms. Even if the mid-bass driver was 90cm above the floor, floor bounce would only increase to 1.77ms.
If I move either of the speakers any further from the side-walls to increase delay to side-wall reflection, my already insufficient 45 degree angular separation at the listening position will be further reduced, increasing treble comb filtering and making imaging stability and width worse.
To achieve 6ms delay for side-wall reflections, the speakers would need to be ~1.4 metres from the side-wall, so to maintain a 2.5 metre speaker separation (for 60 degree separation at a 2.5 metre listening distance) I would need a room that is 5.3 metres wide. My room isn't even 5.3 metres long, let alone 5.3 metres wide.
6-10ms is a fantasy for most people, so basing speaker design assumptions around being able to achieve this seems ludicrous.
Ok, I'll give you that some fortunate people have much larger living rooms (including myself, when I lived in New Zealand where living rooms are generally more North American sized) however I don't see how that gets around the floor bounce.
No matter how big your living room is the elephant in the room is the floor bounce arriving somewhere in the 1.5-2ms range, so how is a true 6ms to first reflection even possible, let alone 10ms ? Nobody seems to pick up on this point.
If the floor bounce is ever present, and a large percentage of people will only ever see side-wall reflections with a maximum of say 3-4ms, then other means are necessary to achieve satisfactory results, and that includes taking floor bounce into account in the speaker design (Allison approach, line arrays, etc) and to provide directivity at higher frequencies.
If you can't increase the delay of side-wall reflections enough, your only other recourse is to reduce their amplitude. You can do that either with side-wall treatment, or speaker directivity.
As far as i remember i mostly listened at about 2m distance and i got
that "faint" and "fuzzy" impression in highs.
It is a "functional" speaker in that version though, but not "definite"
enough for me.
When listening farther "fuzziness" and "vertical stretching" diminishes,
but kind of "faintness" stays.
FR is too ragged and timing too smeared when all 6 drivers are running
fullrange, also i believe the floor reflections in highs are not good either.
(Short initial time delay gap => faintness of the image ?).
Having the virtual sound source like a "balance point" of the three
upper drivers at earheight is preferable.
There is no "wavefront" formed for wavelengths in the order of the
drivers distances. You have to use the part with the smaller (and
nonequal) distances for the highs to get a "statistically well behaved"
radiation.
The equidistant lower part of the driver spacing is for vertical directivity
in bass and midrange ... gaining height with few drivers in that range if
you want.
Last edited:
Good point.
Reducing amplitude by directivity would indeed be the way to go. Like the size requirement for sufficient delay, side-wall treatment is not a reality for most living rooms. Personally, I'd don't really like the looks of it either.
I very much agree but there's also the option of a stereo bipol center speaker which maximizes delay of first reflections. First reflection point could be rigid panels or side walls (although this will decrease the delay slightly):
An externally hosted image should be here but it was not working when we last tested it.
You are not implying that the sound from the reflectors can be considered as the direct sound, am I right?
Nonetheless, getting > 3 m path difference between the sound from the speaker and the sound from the reflectors would still be difficult in this situation.
Anyway, the topic was conventional stereo, so let's stick to that. These other set-ups do sound interesting enough for their own thread, I think. Graaf, maybe you could do the honours of opening one?
The bipol setup I've just shown doesn't rely on reflections, it utilizes them for increased spaciousness. There is a direct component that's why I've drawn a line from the speaker to the listener.
This setup addresses the fundamental problem of stereo speaker crosstalk and head-in-a-vise sweet spot. It also maximizes side wall reflections.
Last edited:
See my answer to Simon.
Why 3m (8.7ms)? I'd prefer >10ms. Linkwitz is fine with 6ms (2.1m).
Anyway, the topic was conventional stereo, so let's stick to that. These other set-ups do sound interesting enough for their own thread, I think. Graaf, maybe you could do the honours of opening one?
You asked "What is the ideal directivity pattern for stereo speakers?". This bipol is a stereo speaker. One of the biggest challanges with it is its directivity pattern. How loud do the side reflections need to be, how to prevent serious lobing over the listening area?
P.S. I prefer staying away from threads opened by graaf.
Yes but the direct component comes from the listeners centre line, so all of the stereo information in the recording must be provided by the wall reflection. Unless you are listening to a mono recording you are relying on those reflections.
You have the added problem that the low frequencies and mono component of the music arrive first, and the stereo part of the image is greatly delayed by the extra path length. I really don't see how this solves any problems and in fact introduces many others.
Fundamental problem ? I wouldn't go quite that far. With sufficient angular separation of speakers I really don't find cross-talk to be a significant problem. As I've mentioned to Dave a few times I find 60-70 degrees more than enough to push comb filtering down to an "insignificant" level where its barely on the threshold of detection.
I also don't think "head in a vice" is as inevitable as some people think. My current set-up is far from optimal in many ways, the room is too small, wall boundaries are too close, angular separation of the speakers is only ~45 degrees, yet I don't have my head in a vice. I can move a foot left or right and still get a pretty well focused phantom image, and even well beyond that I still get a good stereo spread. I've heard plenty of systems where if you move an inch or two the sweet spot is lost.
Oh joy
Precisely what I'm not trying to do. I have more than enough side-wall reflections as it is thank you....
All in all this approach is an extremely poor match for what I'm looking for in sound reproduction. Almost completely the opposite in fact.
Yes of course, that's the idea.
What problems do you see?
Ever listened to binaural recordings? Speaker crosstalk is a fundamental problem. This doesn't negate that there is "good and bad" reproduction within the realm of standard stereo.
I've never heard a standard stereo (60°, equilateral) with a wide sweet spot. Phantom sources always move considerably with head movement of a few inches.
Oh joy
Precisely what I'm not trying to do. I have more than enough side-wall reflections as it is thank you....
All in all this approach is an extremely poor match for what I'm looking for in sound reproduction. Almost completely the opposite in fact.
I can see that in your room the first reflections are essentially prohibiting good sound reproduction (unsymmetrical and too early). The center bipole approach is about adding spaciousness by strong, symmetrical and delayed reflections.
Lets be clear thats not precisely what I say. I say that reflections below 10 ms degrade imaging and IMO should be avoided and that reflections > 20 ms. should be lateral and/or to the rear and are good for adding spaciousness. Since NO reflections < 10 ms. can be achieved in any reasonable sized home room it is a good idea to do two things. 1) to have a speaker directionality and room design that minimizes the Very Early Reflections (VER) and maximizes the later reverberation to the sides and rear. 2) if one can make the first reflection contra-lateral this is better than it being non-contra-lateral (same ear as signal). The brain can deal with the former, but the coloration from the second case takes place outside of the ear and nothing can correct it. I have created a simulation of this difference and I will post it at some point.
These are all very practical recommendations for any small room and the closer one gets to the ideal the better the sound quality.
So lets not misquote my opinions as an impractical "all or none" philosophy.
The brain can deal with the latter too. The issue is arrival time, angle and level. I don't think that the contralateral reflection is any "better". I believe it's worse. The ipsilateral one can be exploited for added spaciousness.
What kind of simulation is that? Would love to hear/see it.
How about massive group delay ? Somehow I don't think having the high frequency Left-Right information being delayed 10's of milliseconds is a good thing...
Yes of course I've listened to binaural recordings, I linked to and discussed some earlier in the thread
Speaker cross talk is a problem when trying to reproduce a raw binaural recording, because the binaural recording has the expectation that there is zero crosstalk between the channels, something that will never be the case in any non-headphone listening.
That doesn't automatically make cross talk between left and right bad in general. It just means that the signal processing and mastering must be different to achieve similar localization cues. So ? If the cross talk results in comb filtering artefacts, as happens at high frequencies when the speaker angular separation is too narrow, that's a problem, yes.
But I would point out that left-right cross talk is normal in our hearing system at low/mid frequencies. At mid frequencies our hearing is predominately checking for amplitude and phase differences between the ears - which can easily be accommodated by stereo speaker placement despite crosstalk.
Except when it results in treble comb filtering, I don't see channel cross-talk as a big bogeyman. Something that needs to be taken into account when recording and mastering, yes, but the vast majority of music already assumes the cross talk is there.
That's a shame. Time to try different speaker topologies and listening arrangements then to find out why ?
I think you missed the part where I said I'm actually getting a pretty good result (which surprises even me) despite the far from ideal conditions. The single biggest success factors in achieving that result in a far from optimal room were directional speakers and toeing in front of the listener.
Yes but I don't want strong delayed reflections...when in this whole thread have I ever asked for that ?
I have never had a listening room where I thought to myself "gee, I wish I had more room reflections". It's always been a struggle to reduce reflections and get slap echo/ RT60 and early specular reflections from the side-walls down to manageable levels without doing very WAF unfriendly changes to the room...
Last edited:
You have to think outside the box
Stereolith (with my modification, The cardboard (TM):
An externally hosted image should be here but it was not working when we last tested it.
Beveridge:
An externally hosted image should be here but it was not working when we last tested it.
Fundamentally the same !
Stereolith creates wider stage, and importantly, all the images are outside room boundaries, the walls really disappear.
- Elias
Please explain how I would achieve 10ms to first reflection in my room which I have previously described. I simply don't believe this assertion, even in the larger rooms I've had before such as 4 metres x 8 metres. Are you also ignoring the floor bounce, which usually arrives well before the side-wall bounce ?
Now if your target is to reduce the amplitude of reflections before 10ms that is an entirely different story and I agree with you, using directivity as well as a bit of diffusion and damping. I definitely try to reduce reflections before 10ms in amplitude, but I see no way to delay the first reflection to beyond 10ms in a typical room, especially if your room has a floor.
Somehow it adds spaciousness
Yes of course I've listened to binaural recordings, I linked to and discussed some earlier in the thread
Speaker cross talk is a problem when trying to reproduce a raw binaural recording, because the binaural recording has the expectation that there is zero crosstalk between the channels, something that will never be the case in any non-headphone listening.
That doesn't automatically make cross talk between left and right bad in general. It just means that the signal processing and mastering must be different to achieve similar localization cues. So ? If the cross talk results in comb filtering artefacts, as happens at high frequencies when the speaker angular separation is too narrow, that's a problem, yes.
But I would point out that left-right cross talk is normal in our hearing system at low/mid frequencies. At mid frequencies our hearing is predominately checking for amplitude and phase differences between the ears - which can easily be accommodated by stereo speaker placement despite crosstalk.
Except when it results in treble comb filtering, I don't see channel cross-talk as a big bogeyman. Something that needs to be taken into account when recording and mastering, yes, but the vast majority of music already assumes the cross talk is there.
I'd say 99.9% of all recordings. Doesn't change the fact that a real center sounds different than a phantom center, especially when moving and or sitting off-center.
That's a shame. Time to try different speaker topologies and listening arrangements then to find out why ?
I think you missed the part where I said I'm actually getting a pretty good result (which surprises even me) despite the far from ideal conditions. The single biggest success factors in achieving that result were directional speakers and toeing in front of the listener.
Yes but I don't want strong delayed reflections...when in this whole thread have I ever asked for that ?
I was a recording and mixing engineer at one time in my professional life and listened to a lot of different speakers in different rooms. I've always had this sweet spot "hypersensitivity".
Don't get me wrong. I don't advocate the bipol center concept. In fact I listen to Geddes speakers every day. The bipol center is just an idea how to add spaciousness without adding extra speakers and/or channels because this is something a RFZ design simply can not provide. Of course if you never heard a really spacious presentation then you will never miss it.
You have to think outside the box
Stereolith (with my modification, The cardboard (TM):
An externally hosted image should be here but it was not working when we last tested it.
Beveridge:
An externally hosted image should be here but it was not working when we last tested it.
Fundamentally the same !
Stereolith creates wider stage, and importantly, all the images are outside room boundaries, the walls really disappear.
- Elias
You were talking about Stereolith and that's this box:
An externally hosted image should be here but it was not working when we last tested it.
If you want to discuss your design, then you should state that unambiguously.
I don't believe your design works the way you've drawn it. Secondly, the level, spectrum and delay of reflections is different between your design and Beveridge. The perception is different.
Now if your target is to reduce the amplitude of reflections before 10ms that is an entirely different story and I agree with you, using directivity as well as a bit of diffusion and damping. I definitely try to reduce reflections before 10ms in amplitude, but I see no way to delay the first reflection to beyond 10ms in a typical room, especially if your room has a floor.
I said "minimize", which to me means "amplitude". I don't know any other way to interpret this word in this context.
And I have said, I don't know how many times now, even in discussions directly with you, that dealing with the floor and ceiling bounces is a real issue for any loudspeaker in a small room and most practically done in the room itself with absorption and diffusion. And in that discussion, you were the one who proposed the "impractical" solutions. You take the practical/impractical sides of the conversation to suite your own argument. I always take the practical side since the impractical is just "blowing in the wind".
I feel like you pay no attention to what I actually say, and only argue about what it is that you want me to be saying.
Markus
I was about to do impulse response measurements in my room and then I thought - those would be misleading since a single microphone makes no differntiation between reflections to contr- or ipsi-lateral ears AND there is a big difference in the two - or so I suspect. So I created a uncorrelated stereo signal with the left signal delayed and summed to the left ear, no change to the right, and the left signal delayed and summed to the right ear, no change to the left. (An idealized situation granted, but interesting none-the-less). I have not had a chance to listen to them, but I will tell you this, in one case there is a BIG difference in the spectral response and in the other case NO difference. It will be amazing to me if these two things sound the same.
Of course, which is better? Well if they do "sound" different and one "measures" better than the other, then by my standards the one that has an unmodified spectral response IS the better one. Otherwise you get back to "accuracy" isn't the goal. "Better" is what you like.
PS. I'd post the signals here, but thats not possible due to their size.
I was about to do impulse response measurements in my room and then I thought - those would be misleading since a single microphone makes no differntiation between reflections to contr- or ipsi-lateral ears AND there is a big difference in the two - or so I suspect. So I created a uncorrelated stereo signal with the left signal delayed and summed to the left ear, no change to the right, and the left signal delayed and summed to the right ear, no change to the left. (An idealized situation granted, but interesting none-the-less). I have not had a chance to listen to them, but I will tell you this, in one case there is a BIG difference in the spectral response and in the other case NO difference. It will be amazing to me if these two things sound the same.
Of course, which is better? Well if they do "sound" different and one "measures" better than the other, then by my standards the one that has an unmodified spectral response IS the better one. Otherwise you get back to "accuracy" isn't the goal. "Better" is what you like.
PS. I'd post the signals here, but thats not possible due to their size.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.