You know the findings better than I do: Is it the diffused image localization in and of itself that listeners prefer; or is it the greater sense of spaciousness and envelopment that usually comes along with it?
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
CD² (Constant Directivity and Critical Distance)
- Thread starter Y~BNA
- Start date
You know the findings better than I do: Is it the diffused image localization in and of itself that listeners prefer; or is it the greater sense of spaciousness and envelopment that usually comes along with it?
That is interesting, and contrary to every test I have ever heard of, participated in or conducted. To my knowledge, precise and particularly stable localization is always preferred, if other parameters stay untouched, and particularly diffusion as a result of wider dispersion and side wall reflections, comes with a deterioration of phantom source localization stability in stereophonic recordings.
May I ask if the following parameters have been evaluated separately? And were only recordings used containing natural, meanungful reverb?
- phantom localization width
- phantom localization stability
- envelopment of phantom sources/ambience
- dept-of-field/proximity
The only correlation I could imagine, is having stable localization and reduced proximity always coupled vs. diffused/distant imaging. Having done quite some tests mainly with recording engineers, (future) acousticians, studio procurement/building executives and musicians, the only thing despite from stable localization, that was consistently preferred, was some ´natural´ depth-of-field, rather distant projection plane, without increased proximity. The moment what you call ´focussed´ localization, was subjectively coming overly close to the listener, being described as ´annoying, obtrusive´ and detached from reverb/envelopment, chances were pretty high this was rated the lowest particularly by musicians and classical recording engineers. Which led to some interesting conclusion that particularly horn-loaded (partly coaxial) speakers, known as being popular as main monitors in the 1970s and 1980s (Urei, Tannoy, JBL, Altec, you name it), were widely dismissed and met hostile reviews.
Could you name an example please? To me it sounds counterintuitive that both the recordings are ´wet´ in the sense of reverb-heavy, and on top diffuse reproduction is preferred. I would conclude this combination leads to overly diffuse and annoyingly distant sound reproduction with reduced clarity, localization and subjective timing.
Interestingly, the audiophile recordings I associate with reverb-heavy, are mostly minimalistic ones capturing natural reverb, not added one. I have been involved in a project with the late T. Nishimura a long time ago, who was praised by audiophiles as the ´Godfather of Denon one-point recordings´. Some were really drowning in reverb, do you mean such?
So you want to say that dipoles were preferred in controlled tests and people like their specific imaging of adding diffuse imaging? I wonder why they are not widespread in the market, if so.
Me thinks, the concept of increasing perceived diffusion, hence deteriorating localization stability in the listening room, was invented and promoted already in the 1960s by Dr. Bose, who built his 901 concept on this philosophy which is an extrem form of a bipole trying to achieve a 8:1 ratio of indirect sound compared to direct sound.
ShadowFiend
Active Member
- Joined
- Dec 20, 2020
- Messages
- 144
- Likes
- 224
Dipole is not widespread enough because
1. They are open baffle or planar, so it is not good as box speaker in terms of packaging/shipping/manufacturing cost.
2. Due to front-back cancellation, it is not a suitable for passive speaker unless you accept the very low sensitivity or very big size woofer/planar.
3. Due to the limited bandwidth of dipole region in each driver and excursion limit, a good dipole speaker need at least three way to achieve the smooth frequency response and directivity on and off axis. So it will not be good choice for entry market product.
4. Dipole speaker needs to place in distance at least 1m (IMHO 1.5m) from front wall to achieve sufficient delay between direct and the most prominent indirect sound
Most of them are business reasons, but business reasons are deal breaking in most case.
Absolutely correct, and very relevant in a sound reinforcement environment with potential listening distances significantly varying. For home listening, not so much of a problem, as both the line source and point source can be optimized for a specific listening distance window. If the crossover freq is in a sensitive band, though, it might require applying different EQ strategy, and indeed that is what I found to be the case with the Perlisten loudspeakers featuring this type of array (which is a line source between approx. 1 and 4K and a single waveguide-loaded tweeter above that).
With stereo systems, it is not always a sharp threshold beyond which the localization collapses suddenly, it pretty much depends on the reverb pattern and from which distance on parasitic localization, like side wall reflections, are deteriorating localization width or stability. As the latter is frequency-dependent, it is not really possible to predict it. What is a pretty common case, though, is loudspeakers with a wide-radiating tweeter in its lowest band of playing alone (above the transitional band between midrange and tweeter) with side-walls closely, oftentimes produce dominant discrete reflections in the 3-6K band, making the sibilants of human voice ´leap or fly in the room´, while midrange localization alone is relatively stable.
Such very rarely happens with localizing real sources, but some room properties such as pretty diffused early reflections, might make the collapse in localization stability also not as sudden as Griesinger is describing it.
I was not aware of his research on how inner phaseshift or group delay are affecting perceived proximity and intelligibility, but it absolutely makes sense and is in line with everything I know from practical experience.
His experiments from Boston Symphony hall, though, are exactly pointing out what I was meaning when mentioning Hamburg and Bayreuth venues (which are both really unique in the world). Removing the early discrete reflections (or diffusing them) are giving a feeling of excellent localization, clarity, transparency and reverb envelopment at the same time.
He is asking an open question which other mechanism are at play defining perceived proximity. From own experiments and some which I have attended, I could suggest several potential explanations. One having to do with the spectral differences between direct sound, early reflections and reverb, how they influence the perception of direction for indirect sound. That is my main point why constant directivity speakers and balanced reverb tonality are of importance if you really want to reproduce a coherent acoustical panorama of instruments and ambience.
He is 100% right, can fully confirm this from practical experience, although he is referring to real source localization and I am doing practical tests to find out about critical listening distance mainly with a dry human voice as a mono source on stereo loudspeakers.
Would go so far to say this critical listening distance for localization stability *must* be found by ear.
A pair of 12" would do:
Not so far off from what many people consider to be a usual tower speaker.
Certainly true, but not much difference from conventional tower speakers requiring a certain minimum distance in order not to get boomy.
To be clear here: I am not speaking in favor of dipole concepts and have a pretty mixed history with setting them up. I found particularly concepts with broad directivity drivers and de facto bipole-pattern radiation in the treble bands, Linkwitz 521 and derivates, to be pretty complicated and unpredictable.
Nevertheless, if these concepts have been proven to be preferred under controlled listening test conditions, or deliver properties which are generally preferred, I would have expected them to get more of a market share.
Why not? I mean, it depends on what exactly gets critical and how you define it.
ShadowFiend
Active Member
- Joined
- Dec 20, 2020
- Messages
- 144
- Likes
- 224
Well I think it is very far, since most people now don't want 0.5-0.7m baffle width. And the exposing magnet is deal breaker for many people also.
Dipole require more front wall distance since the energy it radiated on the back side is much more than box speaker in the region 800-1kHz up.
Lx521 or the likes is complicated but this is necessarily complicated to achieve smooth frequency and constant directivity, and at the same time provide enough sound intensity in low frequency.
Can I ask which dipole speaker did you have experience with?
Your thought is IMHO too simple. Market share depends mostly on entry level products. And it is impossible to make a low cost dipole speaker product to compare with 2 way entry level box speaker in terms of bass output. And bass output is a important factor for customers, accordings to research. Not to mention, the size of dipole speaker is a deal breaking in WAF. With that disadvantage, dipole speakers can only have impact on high price segment, like of Steainway-Lyngdorf, Linkwitz Labs, Kyron, IO design or diy world.
Last edited:
dipole directivity index (at least according to some sources)
Last edited:
So, in your opinion, is the issue with the LX521 an excess of off-axis energy at the lower end of the rear-firing tweeter's range? If so, seems to me one possible "fix" would be to have the rear-firing tweeter's power response complement that of the front-firing tweeter; in other words, for the rear-firing tweeter's power response to "zig" where the front-firing tweeter's power response "zags".
I don't know enough about the LX521 to know whether this is already being done.
Or, is the issue simply that the "backwave" shouldn't have that treble energy to begin with? (Such has not been my limited experience with dipoles and bipoles, but your experience exceeds mine.)
If a redesign of the LX521 was feasible, imo it would make sense to use fairly shallow waveguides on both tweeters to pattern-match with the cone upper-mid, but maybe that's just me... I tend to see waveguides as solutions to a lot of things...
My practical experience with that model is very limited, for example I did not have a chance to go through a practical experiment with figuring out the critical listening distance defined by localization. But I would say that some issue is there in exactly that frequency range, but I am not sure it is really excess off-axis energy between 4 and 8K. A spinorama measurement, or just a simple 90deg frequency response graph (preferably *not* on tweeter plane), might answer this question.
What I noticed is that localization is most stable if the side walls are far away from the speakers, and the listening distance is pretty low, almost nearfield. Not what you would expect from a large open-baffle/dipole model, as textbook dipoles have a cancellation null around 90deg, so side walls should be more or less irrelevant.
A potential explanation would be that cancellation between two sources of inverted polarity, such as the two tweeters, is decreasingly effective the shorter the wavelengths get, if the sources are not at the same point. I mean, 5,000Hz has a wavelength of 6.8cm, meaning half a wavelength expected to cause perfect cancellation, is just 3.4cm, with the error between full cancellation and average net SPL just 1.7cm apart. That is significantly less than the vertical distance between the two tweeters, and at 8,000Hz this only gets worse.
Could work in theory if both tweeters are small compared to the wavelength and positioned very very close to each other. That is not the case with the 521´s tweeters, and I am not sure it is doable with anything but a true one-diaphragm dipole such as a magnetostatic planar.
Not claiming to being able to analyze all potential dipole concepts, but I tend to side with Erik Wiederholtz who kind of rigorously postulated ´not more than 1 tweeter above 5K active´. Don't like tweeter arrays and the kind of combfiltering, phaseyness and matt upper treble they tend to produce, regardless if planar array, multi-dome line source or dipole.
If you want to have balanced tonality of the reverb in the room and the directivity of your concept increases towards higher frequencies contributing to the impression of a ´dull reverb´, you must do something to counter that. I mean, if you were Linkwitzlabs, your midrange and presence bands are anyways creating such second wavefront caused by the dipole, so having a similar pattern in the treble is a logical conclusion if we are aiming for even dispersion and similar radiation pattern.
I just don't think (dipole) cancellation is the right strategy for very short wavelengths.
Excellent idea, but that would basically mean you have to employ two waveguides with a pretty narrow opening angle, so beyond a +-60deg window the treble energy to both the front and the rear is already significantly lower from like 3K upwards. So these waveguides would have to be huge, tweeters would be even further apart from each other, and the aesthetic beauty of the open baffle concept was no more.
Maybe a tall planar dipole with a moderate waveguide to suppress energy radiated to the window around 90deg horizontal, like a magnetostat or AMT, would be easier. As mentioned, I was astonished by the performance of Dieter Fricke´s Ecouton TransAr which did exactly that already some 15 years ago.
- Thread Starter
- #72
Hi all, very happy with everyone's input. Honored, to say he least, as far as adressed to me. Have thoughts. Progressing shaping of the plan. Will like to share soon, but crazy busy and dead tired this week and last. Sorry. Soon.
This is the manufacturers (!) off-axis graph for the PT6816 planar tweeter, FWIW. No additional measuring parameters. But if a 6-unit stack will attain even slightly wider dispersion than this, i'll be a happy camper already. Later.
This is the manufacturers (!) off-axis graph for the PT6816 planar tweeter, FWIW. No additional measuring parameters. But if a 6-unit stack will attain even slightly wider dispersion than this, i'll be a happy camper already. Later.
Last edited:
Frequency response just up to 45deg horizontally does not tell much about overall directivity, particularly with a diaphragm of that size. You can expect it to narrow down above 5K horizontally due to the diaphragm width, which is a bit too early to use is fullrange. If you want to keep lobing and overly increasing directivity index under control, I would assume that a single unit might go up to 7K, but that already requires a pretty clever strategy to implement a slimmer (super-)tweeter without vertical lobing.
Combining it with a curved or truncated line array equipped with identical units below 2K should be doable, so you can keep the directivity constant between 500 and 5,000Hz. Given how tall the diaphragm area is, I would say 3 or 5 units are ideal, with the central one being active to approx 5 to 7K, and the outer ones having some low-pass filtering applied around 2K or a bit lower.
There are examples of this planar used in arrays, for example on diyaudio
https://www.diyaudio.com/community/...ght-line-array-floor-to-ceiling.347227/page-3
Would say it is a very challenging project but also a promising one. Have heard a ton of speakers equipped with such linesource-shaped planar tweeters, and noticed a tendency that these combined with usual midrange cones kind of appeared to be disappointing, while those combining such planars with midrange line arrays or dipoles, were always surprising in a positive way.
Last edited:
Hi, I think max SPL and bandwidth are the two most fundamental goals/requirements. Then directivity as it relates to practical situation in living room/context. The problem is what directivity? For this one needs to understand what they perceive, and what they want to perceive, and what it means to get that in the practical situation, otherwise any nice will do. But directivity goes only so far, room acoustics and positioning, including the listener, are the other factors that make a perception happen in the listener.
Last edited:
Yep, thats the hifi sound for me too.
But, many people seem to prefer the hazy room sound with reduces localization! Difference between involving and relaxing.
Actually, I do too at times, by mood, by recording and so on. It's really empowring to understand you don't have to lock in on either, but have both changeable at will, by changing listening distance. Please read Griesinger Duke already referred to.
Yeah many people seem to prefer the poorly localizing loud early reflection dominated big hazy sound. But, there is another way to make "wider" image: small enough listening triangle so that perceptually early reflections disappear, sharp localization happens, now just position the speakers wide enough apart so that image seems natural "width", what ever that is to you. If acoustics is good, this can be very big sound if it's baked in to the recording, envelopment. If it's dry mono then it's dry mono though.
Sometimes very dry recordings, wild panning and such can make the close listening too annoying. This can be much more comfortable with early reflections dominating perception, while sometimes the involving sound is the preferred.
I claim that listening setup is ideally positioned like so, to enable both perceptual sounds at will so one can just change perception at will by moving the butt a bit. Possibly just lean back for relaxed big sound, or forward to zoom into it. Difference is brain paying attention or not, explained by Griesinger Duke already linked to.
When speaker has suitably rising DI, suitably adjusted toe-in, very good edge diffraction performance, speakers wider apart don't really sound any different (HRTF stuff, same source more to side seems broghter sounding, DI compensates). Enables max flexibility for positioning and toe-in to be able to optimize both perceptions (any listening distance).
Last edited:
Absolutely, this unlocks listening skill! makes all the millisecond, directivity etc. debate is backwards in sense that it's much faster and easier just listen, than setup everything with tape measure. If you set up a system with tape measure, how do you know the sound is good or not? Must be, right? But, if you listen the thing first by moving yourself and actuaööy hearing how perception changes as early reflections relation to direct sound changes. As it appears there is almost on/off switch (Griesinger limit of localization distance)
You can use your own perception as AB switch to all of it, ability to use logic with perception, all those milliseconds don't really matter anymore because you know your perception changes as you can do it by moving a little. Using this shift in perception enables to learn about what too much early reflection means perceptually, what envelopment is like, localization, stage, depth, whatever, you kinda get to hear them just by moving back and forth and using some thought with it, never knowing what milliseconds you have. You actually learn to listen what you like, and how any written concepts on forums and books relate to your perceived reality.
Last edited:
Deviating from an equilateral stereo triangle might be possible in some cases of nearfield listening, but it is far from being an ideal solution, or leading to an imaging of both excellent localization and consistent ambience at the same time.
A certain degree of added reverb in the listening room helps a balanced imaging, if it is sufficiently attenuated, diffused and tonally balanced IMHO.
My perception differs. Having pretty dry recordings tending to show a high level of proximity, is by no means cured by additional early reflections. In contrary, the latter is oftentimes leading not only to blurred but unstable localization, which I find particularly annoying with more or less dry recordings. Again, a certain level of tonally balanced reverb in the listening room does help here. We should not forget that reverb tonality does add a quality of envelopment and perception of direction as well as reduced proximity.
I tend to dispute the rising DI part. Particularly loudspeakers with reduced energy in the room above 2K and increasing directivity index, tend to cause an effect of reduced proximity, with phantom sources sounding overly direct and annoyingly close to the listener. Unfortunately, a lot of speakers employing either waveguides, coaxials or horns, as well as broad bands of lobing at a typical crossover point around 2-3K, tend to cause this effect in my experience.
There are several potential explanations for this. Direct sound in the 2-5K band is anyways dominating in perceived loudness over diffuse sound (according to Zwicker/Fastl), so reducing reverb will make this effect even worse. On the other hand, dominant level in the 0.8-2K band compared to 2-5K, is according to common HRTF models signaling our brain sound coming in from the rear, which will increase the effect of separated perception of reverb and direct sound, hence leading to even dryer, more detached perception of the latter.
In contrary, having a bit more of diffuse sound level in a narrow band somewhere between 2.5K and 4.5K and attenuated reverb in the 0.8-2K region, does add perceived distance and makes the listening room reverb be perceived more as the natural reverb of the dry sources. Stress on ´narrow band´, as a broad-band boost will lead to tonally kinked i.e. overly brillant reverb. Years ago, i have been taking part in an experiment organized by a university institute and a loudspeaker manufacturer, including an A/B comparison of two almost identical monitors, calibrated for identical on-axis-response, with the only difference being the directivity above 2.5K (one had a waveguide in this band, the other one did not). The differences were pretty telling.
done
Similar threads
