Markus - just for you!!
But don't upload here, it will get buried and forgotten just like many of your posts. Upload the PDF to your website, link it from a page on your site so spiders can find it and link it from forums like this. </SEO tips for free>Or the air, don't you see the geographical correlation?
I hope more people will do the test and a pattern emerges.
Yes, I thought of a drum after posting. I did get a bit carried away with my monopole vs dipole in real life analogy...
A piano a dipole sound source ? Where are the opposite sides with identical but out of phase radiation ? I don't get that one. A bipole maybe, but I'm struggling to see a piano as a dipole or a bipole. Just a large device with a unique spectrum from each surface.
There are a few sound sources that are dipolar in nature, but does that mean its ok to reproduce the entire sound mix of a band through a speaker with a dipole response so that every virtual sound source in the mix has a dipole pattern ? I'm not so sure about that...but I'll drop the point, as its just my conjecture.
Ok, but if unnatural lack of spaciousness was the complaint with the dipole in Floyds test, this is presumably due to the 90 degree notch being near the correct angle (with the speakers toed in) to cause a deep notch in the early side-wall reflection to the listener ?
A deep notch that is highly directional and which a constant directivity system doesn't have.
In fact in a narrow room (speakers on the short wall) a 90 degree horizontal CD system would not significantly attenuate the early side-wall reflection if placed at a more typical ~15 degree toe in relative to room sides, as the reflection angle would be within the +/- 45 degree beam width, the level of side-wall reflection would be fairly similar to the "wide" dispersion conventional design, and the direct path signal would be flat.
In a wide room (speaker on the long wall) it might be necessary to use no toe in or even slight toe out to get a similar level of side-wall reflection to the conventional wide dispersion design, but doing so would still give you a flat direct path signal to the listener.
It's not until you start to significantly toe in the 90 degree CD system before side-wall reflections start to get significant attenuation - in the extreme case a 45 degree toe in in corner placement. But they don't have to be used that way.
It's a point I've made before - by simply adjusting toe in you can adjust the response to suite your own preference of side-wall reflection level, (wide ASW vs "dry and revealing") and cope with a variety of different room acoustics, making the speaker very versatile, while with a true wide dispersion design you're stuck with whatever the room gives you.
With the correct toe in (or perhaps toe out) I'm sure a 90 degree CD system would have rated well in Floyds tests, and I see no reason why it would get the same negative reaction as the dipole, and why the dipole results should be extrapolated to a CD design. Until the tests are repeated we can only conjecture though.
Not all dipoles are electrostatics... I wasn't specifically thinking of electrostatics when making this comment, but rather dipoles in general including dynamic driver dipoles like the Orion. In fact I would say most DIY dipoles would be dynamic designs and few would be electrostatics.
Yes I have heard of cabinet stuffing, of course...perhaps I should have been more explicit in what I meant. When most dipole fans compare dipoles (generally dynamic driver types) with "box speakers", I think that the main quality improvement they are really hearing is a lack of "boxy" colouration, both reduced panel resonances and eliminated internal reflections, rather than a dipole radiation pattern itself being some kind of panacea.
Dipole fans (Linkwitz included) try to rationalize the polar pattern as being a key advantage of a dipole, including the 4.8dB DI, but I really think it's just that - a rationalization, and that the freedom from internal reflections is the main advantage which must be offset with the disadvantages of an "oddball" radiation pattern. (Look how fussy a dipole is with room boundary proximity and positioning)
As for stuffing - yes you can almost eliminate internal reflections with the right stuffing in a sealed box, (or midrange cavity) but not on a 2 way bass reflex design, and yes you can brace the hell out of a box to get panel resonances down to a minimum, but it's not easy.
If you compare a typical not very well braced and stuffed box to an open baffle dipole, the dipole will win on lack of internal reflections without a doubt. The lack of internal reflections is clearly audible. It takes a really good box design to get the level of internal reflections and box colouration down to the level of a dipole, (and I suspect not many DIY designs reach that level) which is why I think some people opt to take the easy route and go for a dipole. I'm not saying it can't be done with a "box speaker" design though.
Lets not confuse power response holes with holes in the off axis response at specific angles though, which is what I was talking about.
Yes a CD design with 2nd/4th order crossover and wide enough driver spacing will have a power response hole at the crossover like any other, but this does not mean it will have a hole in the horizontal off axis response at the incident angle of the side-wall reflection. In fact in an ideal CD design it will not.
The "medium dispersion" conventional designs were not as preferred in the test as the "wide dispersion" designs - why ? I suggest that it may be the increased holes in the off axis response at the incident angle of wall reflection.
In other words small holes in the power response don't matter, but holes in the horizontal off axis pattern at the incident angle of wall reflection do matter.
In this regard a CD design should come out on top of even a "wide dispersion" conventional cone and dome design, which will typically still have small dips in the wide off axis response.
There is still an odd contradiction though - many times before you've suggested that spectral flatness of the side-wall reflection is not a significant factor, and that small dips in the power response also don't matter, and yet in those tests the wide dispersion speakers were favoured over the narrower dispersion (but conventional design) speakers which would have had increased holes in both off axis response and power response... I still can't reconcile this contradiction.
I really don't think it's anything to do with the power response curve, that much I agree with you. I think the key factors are the polar pattern shape rather than the specific DI, and the spectral balance of the horizontal off axis response. (Minimizing big holes in the horizontal off axis response, particularly at angles that will give early reflections)
I'm not convinced that a CD system can't give a spacious sound either - if that's what a listener desires, by simply reducing toe in or even toeing out. These are all options for room set-up if desired.
Compared to a cone and dome design, an ideal CD design would allow for a more uniform response across the listening window, which from above you would agree is useful, would allow for a spectrally balanced side-wall reflection regardless of toe in, which you so far don't agree is important, allows for a greater direct to reflected ratio depending on toe in, which may or may not be of benefit depending on the room and listener preferences, but which should have enough adjustability with toe in or toe out to suit most listeners.
Things it doesn't have - flat power response, (which you say doesn't matter, and I agree) 90 degree off axis deep notches, out of phase radiation, holes in the horizontal off axis frequency response.
Last edited:
What is so wrong with measuring T60 of a small room ? Especially if the figure is extrapolated from a T30 measurement, as a lot of measurement software tends to do. The "ideal" range is different to a large room, but I don't see why it's not still useful as one measured property of a small room ?
And yes, of course it varies with frequency, isn't that in itself also a useful piece of information ? Again, most software will graph it vs frequency, so it's not necessary to refer to it as one average figure.
The mono sections of Linkwitz's test signal (which I've used before) don't sound any different to Markus' ones to me, or to my own pink noise sources.
The "fuzzy Stereo" sections are simply two different sources of pink noise that have random phase correlation but are otherwise identical in overall amplitude and spectrum. It's the ideal demonstration of correlated and de-correlated sounds.
It's normal for the stereo (decorrelated) pink noise to sound like it's coming from both speakers individually with nothing in the middle, while the mono sections should appear to come from only the middle.
You'll also notice a complete absence of comb filtering artefacts whilst moving your head sideways on the stereo pink noise sections...
Yes, I thought of a drum after posting. I did get a bit carried away with my monopole vs dipole in real life analogy...
A piano a dipole sound source ? Where are the opposite sides with identical but out of phase radiation ? I don't get that one. A bipole maybe, but I'm struggling to see a piano as a dipole or a bipole. Just a large device with a unique spectrum from each surface.
It must be a dipole at low frequencies. A grand piano has a large sounding board in an H frame dipole configuration, open above and below. The sounding board is a thin spruce diaphragm. This power point doesn't clearly show the dipole radiation (I'm guessing they measured it on a hard floor 5 wall chamber) but gives nice plots of sounding board modes.
https://ccrma.stanford.edu/courses/150/lectures/14_KEYBOARD_INSTRUMENTS.ppt
I agree that the specific reflections will be different between the two polar paterns. Its the old "micro, macro" thing. The question remains whether a generally more directional system sounds drier, nomatter what the particular patter. Yes I am speculating on this one. Dipoles do seem to be an unfortunate choice if we think that sidewall reflections are desirable.
Take a look at the Toole measurements (the 20 ranked speakers of the first big study) and you'll see he averages off axis bands. After 0 and 15 degrees, 30 and 45 degrees are an average, 60 and 75 degrees are another. Finally he gives front hemisphere and full hemisphere averages. The interesting thing is that the power response trend is fairly well seen in the 30-45 degree average and totally revealed in the 60-75 degree average. I'm sure the curve varies from specific point to specific point but the average (of 8 points, I believe) of the lesser groups totally reveals the final trend.
Toole calls the 60-75 group the "early reflections" group, meaning that curve would be most definitive in the initial wall bounces and a strong part of the in-room curve (not that we believe in in-room curves). Yes a CD speaker of flawless design can preserve good response to wide horizontal angles. Thats not the point. The point is that speakers that weren't CD designs, but did have response holes at angles of the wall bounces were top ranked in his tests. I've reattached the sound power measurement, rank ordered by listener preference (see post 723 for others) and would point out that the 1st, 2nd and 4th ranked systems have mediocre sound power (and off axis response) while bottom ranked #16 and #18 out of 20 have what I would guess would be ideal sound power.
Your argument is that: "Yes, but if they had CD designs of flawless off axis performance, they would rise above all others" Again, if off axis response isn't a significant factor then why would a big improvement in it suddenly matter?
Note that this was the conclusion of Lipshitz and Vanderkooy as well, that holes in sound power are "relatively innocuous".
Their paper is well worth looking at in this discussion. They came to multiple conclusions: making the sound power flat by tiping up axial response to counteract rising directivity is bad. Sounds way too bright. Giving a speaker flat power response and flat axial response (as in flat d.i., perfect constant directivity) is not as bad but still sounds too bright. Falling power response is fine, and holes in the power response are relatively innocuous, peaks are not.
If elevating the power response makes a speaker sound overly bright, but reducing it reduces spaciousness, then power response does, in the end, matter. It implies there is a range of acceptable power response curves. I would suggest that the average d.i. of the better ranked systems in the Toole study are more than just the happenstance of typical speaker design, that they have evolved to meet the preferred d.i. trend. That is, for typical listeners in a domestic room of typical acoustics, that a gradually climbing d.i. of about 6dB through mid frequencies and 10dB at the top is an essential part of ideal design. (And of course holes in the trend are acceptable but peaks are not.)
Finally, note that the L&VK tests used a side firing dipole on top of a forward facing monopole. Any energy they artificially added into room power response went straight into sidewall reflections. If they found that a system's power response could acceptably have holes in its spectrum they were directly putting holes into the early lateral reflections.
Flat spectrum reflections may sound like a good idea but it is still an idea not proven.
David S.
I agree with this. The only thing that changes with room size is the frequency at which a statistical approach applies. At high frequencies RT is totally valid in a domestic room. That is the meaning of the Schroeder frequency: the border between the modal region and the diffuse region.
I'd think that even below the Schroeder frequency the notion of rate of sound decay is just as important. It simply becomes impossible to quantify with a single number per freqeuncy.
Note that people measure RT in scale models. At the scaled wavelengths of interest the model size is irrelevant.
David S.
I would contend that there is a clearly objective test that would differentiate this and its simply the rooms impulse response and which technique yielded the lowest levels of energy in the very early period right after the initial impulse, i.e. < 10 ms. Diffusion is certainly better than doing nothing, which is a disaster, I think that we can agree on that, but I believe that absorption will yield the "better" results because the energy is elliminated. If the diffusion can somehow delay the return of this rearward traveling energy beyond 10 - 15 ms. then I would agree with you that it is the better option because it will add to the LEV and not degrade the imaging. But in any situation that I can think of this won't be the case. And that assumes that the diffusion works ideally, that is also seldom the case. I should point that I do use diffusion everywhere else in my rooms so its not that out that I am opposed to the technique, just its use behind the speakers.
See, all science, no mention of "How it sounds to me".
Sine wave sweep?
Would hearing a simple, slow sine wave sweep add any value in addition to using various pink noises?
Back in my audiofool days I used to use pink and other noises to set my soundstage. The more I got music to stage behind, in front of, and around the speakers, thus achieving audio nirvana, the less it sounded like real music.
So I started using a sine wave sweep from 20-20K to place the images between the speakers where it sounded more like music and less like audio tricks being played on my ears.
Would hearing a simple, slow sine wave sweep add any value in addition to using various pink noises?
Back in my audiofool days I used to use pink and other noises to set my soundstage. The more I got music to stage behind, in front of, and around the speakers, thus achieving audio nirvana, the less it sounded like real music.
So I started using a sine wave sweep from 20-20K to place the images between the speakers where it sounded more like music and less like audio tricks being played on my ears.
but how can we know scientifically that the lower the better? and another question arises - better in what sense?
because we know scientifically that a typical listener does not prefer the lowest levels of energy, quite the contrary, with notable exception of some professional listeners suffering from abnormal occupational hypersensitization
but how it sounds to a typical listener is not at all about subjective sounds to me - it is rather part of science, it is objective how it sounds to a typical, normal human, isn't it?
It is discussed in the following two articles:
I'm on the road and cannot do that from here, but that is the plan. I can write it on the road, but not access my website as the admin.
Two big problems. First is that the sound in a small room does not get difuse until after the T60 is actually gone. So basically the field never meets the assumptions used in defining T60 and its measurement. Second is that it is so short that it becomes statistically unstable - different measurements give different results, its not stable. T60 or T30 have lost their appeal even in large rooms as a meaningful metric. They are even less meaningful in a small room.
If audio to you is simply an exercise in engineering, that's OK. It isn't "to me". I still have to listen to the stuff. "How it sounds" is very important "to me." Nothing wrong with that. Making it sound good using good engineering is going to be more reliable than using voodoo, for sure. And the chicken blood all over the listening room is always unpleasant.
But at the end of it all, you have to sit down and listen to it. How does it sound? Until someone posts a good study that shows that listeners can better identify spacial clues with one technique over the other, or have a consistent preference for one over the other, it's all opinion and a matter of taste. So far, we haven't seen the science.
This doesn't make sense to me - how many reflections must a ray go through before it can be considered diffuse ?
Sound travels the same speed in small or large rooms - in the small room a given ray will reflect many more times in its journey in the same time interval than it will in a large room.
So if anything, the reflections in a small room will become diffuse and semi-randomized in their physical orientation and phase in a shorter period of time than a large room.
Does it get diffuse before the RT60 time is up ? Well as I asked, how many reflections do you need ? Imagine a small room like mine 3.45m x 4.8m x 2.4m. Measured RT30 is 0.4 seconds give or take in the midrange/treble, (third image attachment) for a ray to travel along the longest axis of the room takes about 14ms.
So worst case scenario of a reflection primarily along the long axis of the room is that we can still get 20 consecutive reflections into 280ms - well under the RT30 of the room. This is worst case - every other reflection angle will result in many more reflections for the same time period, for example imagine a single ray from the speaker at an angle that first bounces off the floor, then the side-wall, then the ceiling - all of which could happen within 14ms or so, so some radiation directions at odd angles could experience as much as 60 reflections in the same 280ms period, even in an empty shoebox room, let alone one littered with objects that will introduce further diffusion.
As far as I'm concerned, that's plenty of reflections per ray departure angle to become diffuse well under the RT60 time.
I've attached a couple of expanded time and amplitude scale versions of one of the ETC's I presented earlier in the thread - showing a full 70dB range, and 250ms and 500ms respectively. I've highlighted the last individually discernible discrete reflection at around 36ms in red, beyond that all I see even when zooming in is statistical variations, and the overall envelope decline is a perfect straight line from 36ms to nearly 300ms. (Beyond this the fan noise from my computer is setting the noise floor) 90% of the decay time is linear decay without discrete reflections poking their head above the overall reverberation.
I don't see a problem with this as a measure of the decay time of the room, particularly when broken down into different frequency ranges ?
Unstable in what way ? Two identical measurements one after another don't match ? Or two measurements with slight microphone location differences don't match ? Just because the time period is shorter I don't see why the measurement suddenly becomes non-deterministic. I haven't noticed any measurement to measurement variations, and small positional changes of the microphone of a foot or so make almost no difference, providing the mic is sufficiently away from nearby objects. Perhaps if you notice inconsistencies in readings, its the way your software/hardware does the measurements. ARTA uses linear regression to determine the slope of the envelope decay in each frequency band rather than just taking a reading at a specific time delay, which should be fairly resistant to random fluctuations.
Attachments
Last edited:
Lots. Even the sound field in a concert hall isn't perfectly diffuse (homogeneous and isotropic). This is what Toole writes about RT60:
"In a small listening room we are in a transitional sound field, consisting of the direct sound, several strong early reflections, and a much diminished late reflected sound field. What we hear is dominated by the directional char- acteristics of the loudspeakers and the acoustic behavior of the room boundaries at the locations of the strong early reflections. RT reveals nothing of this. As a measure, it is not incorrect; it is just not useful as an indicator of how reproduced music or films will sound. Nevertheless, ex- cessive reflected sound is undesirable, and an RT mea- surement can tell us that we are “in the ballpark,” but so can our ears, or an acoustically aware visual inspection."
From http://www.aes.org/e-lib/download.cfm?ID=13686&name=harman
Simon
Your results look very good, but I suspect that they are the exception rather than the norm. Decays in small rooms are seldom as linear as you are showing.
At any rate, I don't make RT60 measurements and don't really see any reason to, except that they might differentiate my room from a "normal" room, but thats all. There is no "ideal" RT60 in a small room (as there is in a large room), you can never get it very high, so there isn't really a reason to measure it. Just make the small room as reverberant as you can (at HFs) and your done.
And when the reverberation time gets down into the ears integration time? What does that mean?
As I said, my friends who do architectural design seriously discount RT60 for use in anything, and that's in an area where there are no questions about its measurement.
Your results look very good, but I suspect that they are the exception rather than the norm. Decays in small rooms are seldom as linear as you are showing.
At any rate, I don't make RT60 measurements and don't really see any reason to, except that they might differentiate my room from a "normal" room, but thats all. There is no "ideal" RT60 in a small room (as there is in a large room), you can never get it very high, so there isn't really a reason to measure it. Just make the small room as reverberant as you can (at HFs) and your done.
And when the reverberation time gets down into the ears integration time? What does that mean?
As I said, my friends who do architectural design seriously discount RT60 for use in anything, and that's in an area where there are no questions about its measurement.
I think that's missing the point though. Just because the direct field and early reflections dominate the sound in a small room, doesn't mean the diffuse field (reverberation) doesn't exist, isn't heard, or can't be measured.
Just because you're not fully in a reverberant field doesn't somehow invalidate measuring the overall decay of the room in terms of RT60, even if the target value of RT60 is quite different than a concert hall. It's still a measure of how quickly energy decays in the room, and an indicator of how well damped the room is for a given room size, and is a directly applicable measurement to make when making overall damping changes in the room.
Depending on how you contrive speaker directivity and where in the room you put diffusion or absorption, you can reduce early reflections without significantly affecting the overall reverberant field and decay time of the room, or contrarily you can keep early reflections relatively the same in level, but significantly reduce the total reverberation time of the room. The two scenarios sound very different, which is why you can't characterise the room only by RT60, but it's still a useful measure for one aspect of the room.
Early reflections can be measured by looking at the early discrete spikes in the ETC curve, as per earlier in this thread, while the overall decay time of the room can be looked at with the RT60, as measured by the linear regression of the envelope decay of each frequency band.
What would cause the decay to be significantly non-linear in a small room ? Large differences in damping in different parts of the room ?
I usually find most "normal" living rooms too reverberant, not the other way around.
This room I find tolerable but still a bit on the too reverberant side. I aim more for around 0.3s.
I don't think there's any danger of that...however if the decay in the room is too slow it certainly impacts on the perception of the envelope of musical passages, particularly those with high crest factors, and it can also mask low level reverberation and details in the recording. This is what I don't like about a room that's too reverberant, I'd rather have a bit drier response that is more revealing of what's actually in the recording. For some reason I also find an overly reverberant room is a lot more unpleasant and difficult to listen to at high volume levels than a room that is better damped. I always enjoy loud classic rock much better in a well damped room for example.
Last edited:
Think about the advance you have gained by doing this simple experiment ! Now you have the knowledge what is one of the most, if not even the most, important factor affecting stereo phantom imaging.
Next comes to find a solution to this..
Markus ! Can you help us out here?
- Elias
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.