I have been following several threads discussing Kef and Revel speakers over the past few months. In those threads I see many posts calling Revel speakers "wide directivity" designs and Kef speakers "narrow directivity" (either in absolute or relative terms). I would like to know if there is a specific definition for what makes a loudspeaker "wide" or "narrow" where directivity is concerned. For example, should one look for a -6 db point at a certain frequency at a certain angle? My issue is that much of the discussion seems pretty subjective. I've spent the morning comparing speaker measurements on Stereophile and the Soundstage NRC measurements, and it seems to me that, with few exceptions (MBL, for example), most speakers of more or less typical design that measure reasonably well in other ways are quite similar in their off-axis behavior. Are the small differences in these measurements very audible?
This is a good topic to bring up; there aren't concrete definitions for sure. Usually wide vs narrow is used in a comparative sense, buf of course the ultimate wide directivity speaker is a perfect omni, and the ultimate narrow directivity is probably some giant horn or beamforming speaker.
In practice, among most speakers on the market, which are mostly forward-firing designs, the line is blurrier. But it tends to be the case that speakers with deep waveguides or horns will have narrower directivity than those with no waveguides or shallow ones.
To answer your question, I believe that yes, 'small' differences are very much audible, especially if the speakers are well-controlled otherwise.
Directivity does encompass the whole frequency range, but when it comes to the primary effects of soundstage I think most people describe - how 'big' the music sounds, how wide the soundstage is, how much the speakers 'disappear' - I personally tend to focus on the region from roughly 2kHz to 10kHz.
If you have Toole's book, section 6.3 talks about the different timbral and spatial qualities attributable to rooms, but it's also in effect a discussion about loudspeaker directivity, as directivity tells you how a speaker interacts with a room. He presents this chart, the upper portion which describes the estimated frequency ranges responsible for different aspects of spatial perception:
A summary of his descriptions follows, with some interpretation by me.
Direction, Depth, and Distance are pretty straightforward, and are pretty much attributable to rooms and speaker placement. As far as I know, there isn't a way of making a pair of speakers sound like they are very far away through directivity alone.
Envelopment "is the impression of being in a specific acoustical space." Getting a good sense of envelopment to, say, imitate a concert hall, seems to require multichannel listening at home. This part may be what gives giant woofer speakers and cardioid speakers their particular sound, as they will tend to have narrower directivity in this region than smaller speakers which approach omnidirectionality.
Image shift, Image broadening is the interesting bit that it seems we mostly associate with 'soundstage' -- or at least, I do. Here's how Toole describes it:
"Image size and position. Strong reflections have the ability to slightly shift the apparent position of a source in the direction of the reflection and/ or to make the source appear larger. In live classical performances this is called ASW (apparent source width), and audiences like it. In sound reproduction there is evidence that the tendency continues. Small positional shifts are innocuous because nobody knows what the intended location really was. Image broadening is what happens in live performances— the “air” around the instruments and voices. It gives an acoustical setting for the sound, rather than having it float as an abstract pinpoint in space— although it seems that some listeners like that enhanced effect."
I came to the conclusion of focusing on 2Khz to 10kHz for soundstage on my own before I'd really noticed this image, in which Toole defines roughly 1k to 8k is most important for image broadening effects (and again, these are estimates).
So to your point about narrow vs wide and what to look for, it can be hard to compare exactly because different measurement sources present directivity metrics in different ways. I personally like to see roughly how many dB down a speaker is from the direct sound at a few frequencies between 2kHz and 10khz. For example, here's the revel Salon 2, a very well controlled speaker with very wide directivity for a monopole:
From on-axis to 75-degrees off axis at, say, 7kHz, it is down only about 7-8 dB.
Compare that to the Klipsch R-820F, another speaker with good directivity, but much narrower:
It is down about
19dB in the same region. That means that any sidewall reflections from this angle will be way quieter, and when they perceptually combine with the direct sound, they will do much less image broadening. You're more likely to hear the sound as coming directly from the speakers, especially with hard-panned music, but on the other hand positioning may appear sharper/more pinpoint for the space in between, which some people like .
Most speaker comparisons are not quite as different as the above but I believe differences in directivity of even a couple of dB are audible.
This is actually pretty intuitive though if we take these facts together. As a thought experiment, imagine you tried to create a 360-degree speaker setup in an anechoic chamber. Your mains play a sound at 80dB. Then you play a slightly delayed sound from your 'side' speakers at 74dB, and then again at 70dB.
Ignoring perception of volume (you'd want to perceptually level match), it should be no surprise that 74dB playback should sound quite different from the 70dB playback in the
spatial domain. The net effect should be that the 74dB playback sounds 'bigger.' The precedence effect tells us you'll still hear the sounds as coming from the direction of the mains, but the apparent source width gets larger due to the imitated reflected sound. You could probably hear the difference of just 1 or 2dB as well (there are actual studies that essentially operate in this fashion but I do not remember the specifics).
TL;DR Small directivity differences should be audible given similar performance otherwise. Most well-controlled speakers with no waveguide or shallow waveguides are wider directivity, while waveguides and horns will tend to narrow directivity in the key frequency range.
Where things get more complicated is when evaluating a speaker with uneven directivity, as it will be narrower in some parts and wider in others.
