Is constant directivity sufficiently beneficial to justify distorting the direct sound? Compared to the non-CD speaker that maintains an undistorted direct sound and smoothly and progressively beams the high frequencies?
Andy,
The directivity of a speaker has no direct relationship to distortion.
Do you have polar plots of a 6.5" or larger loudspeaker that "smoothly and progressively beams the high frequencies"?
Art
Like an omnidirectional speaker, Earl needs to roll off the high frequencies of the direct sound of his constant directivity speakers in order to get a reasonable rolled off steady response. Rolling off the high frequencies is distorting the input signal which has a flat frequency response.
When you listen to music the direct sound is heard undistorted and the radiation pattern of the musical instrument plus room rolls off the high frequencies of the indirect sound to give a rolled off steady response. Now a loudspeaker isn't a musical instrument but having to distort the direct sound which the brain pays a lot of attention to is less than ideal. Perhaps it is a good compromise in order to gain the benefits of constant directivity but perhaps it is an indicator that what is going on with the indirect sound could be improved. Hence my question to Earl about the benefits of constant directivity.
Am I allowed to use a set of drivers of varying sizes, waveguides and crossovers? In which case I would point you at reviews of a number of competent speakers like those mentioned earlier and Revels. Although I do wonder about that shallow waveguide. Does anyone know why they do that and not use a deeper one?
The shallow Revel waveguides typically lose control at 10k or so and beam strongly. It seems fairly safe to assume Revel intended to do this and I was wondering why? Or perhaps it is the lesser of two evils?
Don't you think a slightly deeper waveguide would look better?
I don't think it is sufficiently beneficial. The direct sound is far more critical.
Lets remember that we don't hear the d.i. curve. It describes a relationship between the axial response (direct sound) and spherically averaged "power response". As such we can have any frequency response we want or any power response we want with a givin d.i. curve. We just can't have both.
Rather than directivity index, what we do hear is the reverberant field in the room, a combination of power response, room acoustics (alpha vs. frequency) and listening distance. That shows the difficulty of defining an ideal d.i. curve: it is only one variable of several.
If its power response we are talking about, a number of studies of the importance of power response have been done. The most direct is Tooles's orriginal study in rank ordering speakers. He found a very high correlation of listener preference to flat and smooth axial response. He found no corelation with power response. Actually there was a negative correlation as some of the poorly ranked speakers had the flat and smooth power response.
Lipshitz and Vanderkooy did tests where they manipulated axial and power response independently and came to a number of conclusions. Achieving flat power response by rising axial response sounded way too bright. Achieving flat power response with flat axial response was also too bright (a true CD case). They found that large dips in power response were hard to detect but that peaks might be detected. (My tests are also corroborating that.)
The Sean Olive tests showed a strong correlation between axial response flatness and smoothness and listener preference but no particular power response shape was preferred.
The inference is that rising d.i. is better than flat d.i. but no exact d.i. curve can be defined. Also, axial response must be given greater importance than power response. It might be possible that a falling off axial response could compensate for a brighter power response (flat CD) but that would have to be room dependent. In a deader room that would clearly lead to a dull sounding system.
Finally, when most people speak of constant directivity they are referring to a system that is constant directivity only in some range. A CD horn with a 12" or 15" woofer has climbing response for more than half its frequency range. This tends to water down some of the supposed benefits of truly flat CD, but perhaps it pushes them towards the desired norm of performance.
David
Many thanks for the nice summary David. Perhaps what is required are speakers with mechanical and/or DSP directivity controls.
You can only limit the dispersion of the sound wave, not increase it by using a wave guide. Beaming is related to the circumpherence of the radiating surface (horn driver/ horn mouth), and beyond a certain point, the horn driver will start to beam right through the wave guide without being influenced by it.
Even Revel can't change the laws of nature, so it is most likely not on purpose that this happens.
The listening axis response is the most critical, but nothing says that it has to be flat with frequency. No tests have shown that (smooth yes). As Dave says "we do hear the reverberant field in the room" and this is more true for a lively room than a dead one. If we want the reverberant field to have a neutral sound with respect to the axial sound then the power response must follow the same general shape as the axial response because the power response is highly correlated with the rooms reverberant response unless there is a huge frequency variation in the absorption, but then again I am assuming a fairly lively room here (better for spaciousness) so there isn't a lot of absorption. This means that a flat DI is required. It is really quite simple.
Clearly in a dead room only the direct axial response matters and DI is completely irrelevant, but those kid of rooms are not really very good listening rooms.
This all ignores the timing of the reflections which is also critical for imaging. The fewer very early reflections the more precise the phantom imaging will be. This favors a high DI, again in a fairly reverberant room, because in a dead room there is no DI preference, no very early reflections and no spaciousness.
Clearly in a dead room only the direct axial response matters and DI is completely irrelevant, but those kid of rooms are not really very good listening rooms.
This all ignores the timing of the reflections which is also critical for imaging. The fewer very early reflections the more precise the phantom imaging will be. This favors a high DI, again in a fairly reverberant room, because in a dead room there is no DI preference, no very early reflections and no spaciousness.
This is not true and there are plenty of studies that contravene it. As mentioned the L & V studies show that having flat axial response and flat power response will give a response perceived as bright. The Toole studies and the Olive studies both show that flat axial response is required and power response is not a key factor. (Short of possibly revealing resonances)
We have to separate whether we are talking about perceived frequency response or perception of spaciousness. We hear the reverberant field level for spaciousness. That the steady state room response and radiated power are correlated has nothing to do with perception. This is the mistake all too commonly made: "I measure this frequency response in the room, therefore I must hear it." Even Toole, who tends to like in-room measurements, said last year in a talk: "we mustn't make the mistake of comparing two ears and a brain with an Omni mike and a spectrum analyzer".
Steady state response measurements have been misleading us for decades. This is why every DSP room correction designer has to come up with this weeks favorite "target curve" and also why the cinema folks have such a mess with 5 versions of the "X curve".
The point is that if non-flat response is made more palatable by relatively brighter power response, then you will be more dependent on room acoustics to define you perceived frequency balance. It seems much surer over a range of rooms to have a flat early field and declining power response, as most (all?) well ranked loudspeakers appear to have.
I'm not disputing the spaciousness issues, but more the perception of frequency balance.
David
These two studies then contradict one another. If power response is not a factor then the L & V studies should not have concluded that flat power and flat axial will sound bright.
Further, why do Toole and Olive even show power response and DI if it is meaningless?
Other than that I am not sure what it is that you are saying. We seem to be going around in circles.
I would tend to agree with this except for the contradiction as noted above. A CD speaker with flat axial response will sound bright. So you turn down the treble a little and everything is copasetic.
Last edited:
Except you can't really separate them. The D/R ratio, which does depend on DI + room acoustics, will influence the perceived frequency balance as well as spaciousness AND localization in a room.
To my ears, balancing the D/R ratio seems very important. Too high and it sounds dry and bright (maybe the reason why flat axial + flat power sounds bright was because the D/R ratio was high as a consequence of high DI ? No evidence for that though.). Too low and you loose localization and a certain sense of "presence".
This is probably the reason why I like dipoles: they're somewhat in between "classical" flat-on-axis declining power monopole loudspeakers and highly directional horn or large woofer + waveguide combinations. Also, they maintain directivity down to 100-200 Hz, preserving a higher D/R, as opposed to transitioning to omni and lower D/R.
May want to rethink that.
Allright then. I said:
"The D/R ratio, which does depend on DI + room acoustics"
The D/R ratio at a certain distance is linked to the directivity of the source (loudspeaker). Assuming uniform room absorption, one can define a "reverberation distance", the distance at which the ratio is 1, and that distance depends on the DI. See for example Linkwitz: Room Acoustics
If room acoustics/absorption are not uniform (and they never are) then that will also play a part. Which is why I said "+ room acoustics".
"will influence the perceived frequency balance"
We did conclude that at least peaks in the power response should be audible via the reverberant field as colorations, didn't we ? The whole discussion so far was around frequency balance.
"as well as spaciousness"
This one should be quite obvious. Spaciousness is determined by the reverberant field, hence D/R ratio again.
" AND localization in a room."
Griesinger found out that good localization, be it of real sources or phantom images, is correlated to maintaining a minimum D/R threshold at various frequencies. See:
http://www.davidgriesinger.com/direct_sound.doc
http://www.davidgriesinger.com/spatialization_and_loc.doc
The first of Griesinger's papers above is strong support for my contention that high DI speakers lead to better imaging. It is the delay that is created between the direct sound and the onset of reverberation that accounts for this. Low DI speakers will have a small delay and high DI speakers will have a larger delay. This is not a trivial thing.
You can't compare the situation in a typical living room with that of even a small hall. The small delay in a living room will be 'absorbed' by the Haas effect.
What Griessinger is talking about is reverberation. And don't forget, he is talking about the halls, not the sound sources. Virtually all naturally occuring sound sources, including talking/singing heads, radiate mostly in 4pi. No 'windowed' controlled directivity there.
The only delay you are ever going to buy with directional speakers is a reduction of the adjacent boundary bounces. Once the directional sound hits the far side it is bouncing around and part of the diffuse field.
The amount of reduction you can expect is only the d.i. difference between wide dispersion and narrow dispersion, at most 6dB. I think a few judiciously placed absorbers can do much better than that. They can also achieve it in a deader room, which will be much more comfortable for conversation and general living.
Do you have any impulse responses or energy time curves to support this?
David
"very early" is undefined here (and the level relative to direct sound goes unmentioned)
No, it isn’t. SL’s experiments with the rear tweeter ORION demonstrated that. We perceive the treble rolloff of the direct sound as dull or “distant” even if the overall balance sounds “better”. Rolling off the reverberant field while maintaining the direct field as flat is what “works” with most recordings . . . and it’s why conventional box speakers with beaming highs manage to sound as good as they do in “normal” listening environments.
With my dipoles (about 4 ft from the front wall) the first perceivable reflection (apart from floor bounce) is from the front wall/ceiling junction . . . the resulting delay is about 13 ms., the next is from the front wall corners, delayed even more. Those reflections increase perceived "brightness" (if not suppressed, either by rolling off rear radiation or suppressing the reflections), but do not seem to compromise "imaging" at all. The overall enhancement of the reverberant field does, however, give enhanced "spaciousness" (compared to highly directional forward firing speakers with the same "on-axis" response placed in the same location).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.