Nice article on correlating speaker measurements with sound

[Defcon]

At listening distances greater than five or six feet, more sound power is delivered to the listener's ears by the reflections than by the direct sound, for most speakers in most rooms. I believe the audible effects of this reflected energy are significant (which is why there's a room treatment industry, as you pointed out), so I agree with your concern with what's happening off-axis. Toole found that not only the direct sound matters, but also the first reflections, power response (sum of all reflections), and directivity index (how directional the speaker is in general).

Correct, off axis response is very important. Also listeners subjectively rate reflected sound which seems to envelop them as better sounding. First order reflections actually help the sound, as does not having a sweet spot.

 

[DukeL]

Also listeners subjectively rate reflected sound which seems to envelop them as better sounding. First order reflections actually help the sound, as does not having a sweet spot.

The spectral content, timing, density, direction, and decay properties of reflections all matter. For instance, a spectrally perfect reflection that comes from the same direction as the first-arrival sound tends to be interpreted as coloration, especially if the delay interval is fairly short, but a spectrally slightly imperfect reflection that arrives from a significantly different direction after a generous time delay (perhaps 10 ms or more) is perceived as enriching both timbre and envelopment! This to me is one of the more interesting aspects of psychoacoustics, as it seems to invite taking another look at how speakers should interact with rooms.

Would you mind telling me what you mean by this phrase: "...as does not having a sweet spot"?

Thanks!

 

[Defcon]

The spectral content, timing, density, direction, and decay properties of reflections all matter. For instance, a spectrally perfect reflection that comes from the same direction as the first-arrival sound tends to be interpreted as coloration, especially if the delay interval is fairly short, but a spectrally slightly imperfect reflection that arrives from a significantly different direction after a generous time delay (perhaps 10 ms or more) is perceived as enriching both timbre and envelopment! This to me is one of the more interesting aspects of psychoacoustics, as it seems to invite taking another look at how speakers should interact with rooms.

Would you mind telling me what you mean by this phrase: "...as does not having a sweet spot"?

Thanks!

I meant that from what I've read, speakers with good off-axis response have no real sweet spot, or more accurately its very wide, which is better for listeners. I have also read speaker reviews which talk about how with some speakers its important to sit in the sweet spot, do proper toe-in etc, all of which is fine but not really achievable in a home environment for most people and esp for multiple people.

From my limited knowledge, this is quite a complex subject - e.g. what kind of room treatments are needed, should they eliminate all reflections? What kind of room correction takes into account reflections?

Also current room correction software is all extremely complex and not understood, you can run Audyssey etc but most people then advise fine tuning it, running REW etc, that is way too complex for people.

 

[DukeL]

I meant that from what I've read, speakers with good off-axis response have no real sweet spot, or more accurately its very wide, which is better for listeners.

Thanks for the explanation! That's what I thought you meant, but didn't want to jump to conclusions.

And yes, I much prefer speakers with a "sweet blob" listening area rather than a narrow "sweet spot".

There is a particular type of off-axis response pattern that is especially good at this. First, let me give some background:

The ear localizes sound by two mechanisms: Arrival time, and intensity (or loudness). If the sound from the two speakers arrives at the same time, the image will be pulled toward whichever speaker is louder. And if the loudness is the same from both speakers, the image will be pulled towards whichever one arrives first. And if they're both the same, you have a strong center image.

So if you have "normal" speakers firing straight ahead, all is well in the middle of the sweet spot, but for someone off to the side, the near speaker "wins" arrival time (obviously) but it also wins intensity, in part because you're a bit closer but also because you're more on-axis of that speaker, where the upper mids and highs are going to be the strongest, and it's the upper mids and highs that convey most of the localization cues. As a result, the center image often moves over to the side farther than the listener has, because both mechanisms support image shift to the near speaker. This is why center-channel speakers are needed in most home theater systems - they're not so much for the person in the sweet spot, as for the people who aren't.

If our speakers have an unusually wide and uniform radiation pattern, there may be little or no discrepancy between the loudness of the near and far speaker for off-centerline listeners, but the near speaker still wins arrival time. So the image still shifts, but not as much.

Now what if there was a way to get one speaker to "win" arrival time, and the other to "win" intensity? The two localization mechanisms would partially cancel one another out, and we'd still get a decent soundstage from well off to the side of the central "sweet spot". Well, it just so happens that there is indeed a technique for doing exactly this. I learned it from Earl Geddes. Here is what we need:

Our wide-sweet-spot speaker's off-axis response should fall off smoothly and fairly rapidly, such that it's -6 dB at 45 degrees off-axis (we's say this pattern is "90 degrees wide") across the top part of the spectrum (ideally from 700 Hz on up but from 1500 Hz on up is still okay). Next, we toe them in aggressively, at maybe a 45 degree angle, such that their axes criss-cross in front of the listening position (and we've voiced them with this configuration in mind). So for a listener off to one side, the near speaker naturally wins arrival time, but the far speaker wins intensity! This is because the off-centerline listener is on-axis (or very nearly so) of the far speaker, but well off-axis of the near speaker.

The key to this working well is, the near speaker's off-axis sound must fall off rapidly and smoothly. This cannot be accomplished with domes and small cones - we need to use a 90-degree-pattern "constant-directivity" waveguide or horn, and cross over to the woofer where its pattern has narrowed to 90 degrees. A 10" mid woofer has a roughly 90 degree wide (6 dB limits) pattern at about 1.6 kHz. That's close enough for us to work with.

There are technical challenges to making this work of course, but they are not insurmountable.

I have several customers who bought this type of speaker and subsequently sold their center channel speakers. So apparently it can work pretty well.

From my limited knowledge, this is quite a complex subject - e.g. what kind of room treatments are needed, should they eliminate all reflections?

Having experienced what a real professional can do as far as room treatment, I no longer do much "armchair quarterbacking". Room treatment is kinda like crossover design - you want to use the right commponents in the right values in the right places. I suggest Jeff Hedback of Hedback Designed Acoustics (a multi-award-winning studio designer who is still affordable), or the professional of your choice. Getting smooth decay across the spectrum is not easy to do, and a few hundred dollars spent on a professional will work much better than spending that much more on acoustic treatments that aren't the thing you most need.

I remember well my first time in a room Jeff had specified the acoustic treatment for. The room was rich and lively-sounding, and had this unusual characteristic: When you closed your eyes it felt like it was about twice as big as its actual dimensions, and that translated over to music playback in the room.

That being said, in general I prefer a rich, well-energized, lively, and highly diffuse soundfield - in some ways almost the opposite of the overdamped rooms we can get from installing a lot of acoustic foam without really knowing what we're doing.

What kind of room correction takes into account reflections?

I don't like to use room correction anywhere above the bass region. If your speakers have a radiation pattern problem (like an off-axis energy flare at the bottom of the tweeter's range, 2-4 kHz, where the ear is most sensitive), that cannot be fixed with EQ. That's an acoustic problem that can only be fixed at the loudspeaker design stage. If we EQ the room response of this speaker, now we have a dip in the direct sound in that region, and our upper harmonic richness and sense of texture are lost.

Maybe this is overly simplistic, but a grand piano will sound great in almost any room. Yes it will sound best in a good recital hall but probably still really good even in your kitchen as long as you don't bang on it too hard, and it will never outright suck. This is because the reflections aren't a problem that has to be fixed - they are spectrally correct to begin with. Imo that's the way a speaker should behave too.

 

[mtrycrafts]

The solution to this needs to come from the manufacturers, and Floyd/Olive have said as much. Speaker companies need to be a lot more transparent and publish FRs and other statistics which are measured in a standardized way. Harmon does this and I admire them for it very much, they are also one of the only companies to use proper testing procedures and DBTs to evaluate speakers.

e.g. some subwoofer companies publish CEA specs which let consumers compare the performance. For speakers, there's really no way for a consumer to make a decision except for flowery marketing buzzwords on the company's site and opinions of other users.

Imagine if other components were sold this way - cars would have no mpg figures, cpu's wouldn't quote their speed etc.

But, but but, audio is different. Let your ears tell you.