Best Loudspeakers for Rich Timbre?

What generates rich timbre out in the real world?

Well, what makes your voice sound richer in the shower?

What makes a grand piano’s timbre and texture so rich and lush in a good recital hall?

And what makes the difference between a good seat and a crappy seat in a concert hall?

The reverberant field.

Get the reverberant field right, without screwing up the first-arrival sound, and you will have rich timbre. The best speakers for that may differ from one room to the next. But unless you listen nearfield, most of the sound that reaches your ears is reverberant sound. You don’t get directional cues from the reverberant sound because of the precedence effect, but the reverberant field plays a major if not dominant role in just about everything else.

Exactly what is involved in "getting the reverberant field right" is a big topic and well beyond the scope of this post, but awareness that the reverberant field matters is a crucial first step. It is not the only thing that matters, but it is one of the more important ones, especially if rich and natural-sounding timbre is a high priority.

Duke

dealer/manufacturer

Thank you, cd318.

Like you, I am "inclined to believe that a great speaker should still sound good in any room." My standard example is a grand piano: Sure it will sound best in a good recital hall, but it won’t suck in your kitchen unless you play it too loud. Imo this is because the piano’s reflections are spectrally correct, so their contributions are almost always beneficial (though they can overwhelm in your kitchen if you play too loud).

You also said, "How much life/ reflected sound you want in a room will always be a matter of personal choice." I am currently working on a design that allows adjustments to the reverberant field independent of the first-arrival sound.

Here are some of the general principles I try to follow, as far as the reverberant field and its effect on timbre:

1. The reverberant field’s spectral balance should be a good match for the first-arrival sound. It will have a little bit less high frequency energy because of absorption, but not by as much as is typically caused by beaming. When there is a significant spectral discrepancy between the first-arrival sound and the ensuing reflections, those reflections do not enhance the timbre very much. They can even cause listening fatigue.

2. Early reflections are undesirable, but late reflections are beneficial, provided they are spectrally correct. In general reflections arriving within 10 milliseconds of the direct sound tend to degrade clarity (though they can still enrich the timbre), and in general reflections arriving later than 10 milliseconds enrich the timbre without degrading the clarity. In fact they can actually improve the clarity by giving the ear multiple "looks" at complex sounds (assuming the reflections are spectrally correct).

3. There can be either too much or too little energy in the reverberant field. Too much and clarity and imaging are degraded; too little and timbre is degraded.

For some anecdotal evidence of the above, consider Maggies. Maggies have a spectrally-correct backwave, and when you position them far enough out into the room (five feet gives about 10 milliseconds of delay on that backwave reflection), that’s when the magic happens. Imo Maggies might generate a little bit more reverberant energy than is absolutely ideal, but if you sit pretty close to them the ratio of direct to reflected energy is increased, and clarity is improved.

@cd318, if you will be at the Rocky Mountain Audio Fest in early October, please stop by Room 3002 in the Tower. We’re going to try to make the room seem bigger than it actually is by manipulating the reverberant field. We think timbre will also benefit along the way.

Duke

@prof, my comments about “too much or too little” reverberant energy arise from some experiments in varying the amount of reverberant energy while leaving the first-arrival sound untouched.

Briefly we used a fairly directional main array aimed at the listening area for the first-arrival sound, and then used a dedicated reverberant-field-only array (optimized for good power response) aimed off in a different direction, so that it made no contribution to the first-arrival sound. By varying the level of the reverberant-field-only array we could independently manipulate the amount of energy in the reverberant field. We found that the additional reverberant energy generally improved timbre, but too much and clarity started to degrade. Hence my comments.

@cd318 wrote: “Perhaps @audiokinesis has developed some form of refined DSP (unlike those crude ones found on many surround sound systems) which can subtly change the sound of your room. Perhaps it is possible to increase ’sympathetic’ room reflections which could give a sense of increased realism. Sounds phenomenally difficult but it’s certainly interesting.”

No DSP, but your “sympathetic room reflections” guess is right on the money!

We borrowed a page from concert hall psychoacoustics: The difference between a good seat and a poor seat in a concert hall arises from the amount of time between the first-arrival sound and the onset of reflections. In a good seat, there is a time gap between the two. In a poor seat there is no such time gap; the reflections start arriving too early, and the effect is, clarity is degraded.

These principles can be transported into the listening room, despite the difference in scale. Done “right” (see my reply to prof at the top of this post), we think we can improve timbre with no detriment to clarity. We also think we can reduce “small room signature” such that you actually hear less of your room and more of the acoustic space on the recording.

Admittedly it is highly counter-intuitive to think that ADDING reflections can REDUCE your room’s signature, so let me explain the theory behind this:

The basic premise is, if your system was in a big room, you would hear more of what’s on the recording because your room’s acoustic signature would be less intrusive.

The ear/brain system judges room size by the “center of gravity” of the reverberant energy. The later in time that “center of gravity” occurs, the larger the apparent room size. By injecting additional late-onset reverberant energy, we can push that center of gravity to later in time and thereby increase the apparent room size. By making sure this additional reverberant energy is spectrally correct and by not injecting too much, we preserve clarity.

I’ve been working with improving timbre by paying attention to the reverberant field for many years. The increase in apparent room size (thereby reducing small room signature) is an unanticipated but welcome side effect.

Duke

@fleschler wrote:

"I swapped my front row ticket for rows 10 through 20 to get the best sound for a combination of direct and long reflective sound."

Duke replies:

[capslock][bold][italics][giant font]YESSSSS!!![/capslock][/bold][/italics][/giant font]

Obviously we can’t get those long reflections paths in our living rooms, but the same psychoacoustic principles are applicable... and imo in many cases offer us a window of opportunity to make a worthwhile improvement.

@prof wrote: "I’ve never heard more room sound contribute to more accurate timbre."

Thanks for replying!

Many speakers do not generate beneficial room sound.

Most speakers’ off-axis response has a significantly different spectral balance from the direct sound, and as a result the reverberant energy is not spectrally correct. This can degrade timbre, and can even cause listening fatigue. I can explain the latter if you would like.

As an example of what a spectrally-correct reverberant field can contribute, imagine listening to a grand piano outdoors vs in a good recital hall. The timbre is improved (enriched, we might say) by the spectrally-correct reverberant energy in the recital hall.

So I’m not an advocate of reflections in general, but I am an advocate of reflections done right. Imo "reflections done right" involves paying attention to the spectral balance, arrival time, total energy, and even arrival direction of the reflections, in particular the first few.

Duke

Hi Prof,

Again, thanks for replying.

Imo the size disparity between a 6" midrange cone and a 1" tweeter dome make it impossible for such a speaker’s off-axis response to have the same spectral balance as its on-axis response. At the crossover frequency, the cone will be beaming somewhat but the tweeter’s pattern will be about 180 degrees wide (constrained by the baffle itself), and this wide pattern will hold up for another octave or so higher before it starts to narrow appreciably. Such a speaker’s power response (summed omnidirectional response) can be smooth OR its on-axis response can be smooth, but not BOTH at the same time. And a significant discrepancy between the two is not conducive to good timbre in my opinion.

Omni or quasi-omni speakers tend to generate spectrally-correct reverberant fields, but they also (by definition) send a lot of energy towards the nearby walls. So they tend to generate a lot of early reflections. Imo even spectrally-correct early-arrival energy can be detrimental to clarity, as shown by the negative effects of early reflections at "bad" seats in concert halls. This may be somewhat offset by the extra early-arrival energy becoming extra late-arrival energy after a few bounces.

I believe that I am at odds with Floyd Toole here - I believe he finds early reflections to be beneficial, assuming they are spectrally correct. My source on the negative effects of early reflections in concert halls is David Griesinger, and it is my opinion that that applies to our listening rooms as well.

Duke

@prof wrote: "I don’t know what speaker you are describing in your last response to me. The Waveform monitor has a 5" woofer - both it and the Mach Solo measured quite well."

Duke replies: I couldn’t find the diameter of the Waveform midrange driver in a quick search, so I just decided to make a generic comment. A 5" midwoofer probably has a 3.5" cone diameter, so there may still be a significant off-axis discontinuity in the crossover region because the tweeter’s pattern may be considerably wider than the mid’s.

@prof: "speaker designers... get fairly pig-headed about the path they’ve chosen..."

Duke replies: Grunt grunt. Oink.

@prof wrote: "But when I auditioned [the DeVore O/96] several times against a bunch of more "neutral" speakers, sure some of the defects were likely there in the mix, but not remotely to the overriding audibility the nay-sayers make you fear, and to my ears they were doing SOMETHING really wonderful that most of the other speakers weren’t. (A certain combination of organic tone and body to the sound)."

Excellent description of "what matters most". While the specifics of "what matters most" may change from one listener to another (and from one designer to another), imo you nailed the essence, which is these two things:

1. A speaker must do SOMETHING so well you can get lost in the music. That something can be timbre, imaging, coherence, slam, PRAT, low-level detail, whatever. But it must do something wonderful.

That’s the easy part.

2. The HARD PART is, the speaker must not also do something so poorly as to ruin the magic and collapse the illusion that its "something wonderful" just created. There are more things that can go wrong than I can begin to list.

Apparently the DeVore O/96’s indeed do their something wonderful and then don’t turn around and do something so poorly as to destroy the illusion. Imo that’s the magic formula, and it’s much easier said than done. Kudos to John DeVore.

As for "accuracy", one of the worst-sounding prototypes I ever made was the one with the flattest response. As I tweaked the design closer and closer to flat, it sounded worse and worse. I pressed on, having faith that the heavens would open once I had achieved flatness. Nope. These days my target curve for home audio slopes gently downward as we go up in frequency, so I guess I don't even try to build objectively "accurate" speakers.

Duke

@fleshler mentioned his "10 foot wide good sound criteria."

There is an unorthodox technique which can give you good tonal balance over a wide area, and even pretty good soundstaging, though the soundstaging will still be best up and down the centerline.

The basis is this principle of psychoacoustics (which I’m going to simplify a bit): The ear/brain system localizes sound by two mechanisms: Arrival time and intensity. We can take advantage of this to still get a decent soundstage even if we are well off to one side of the centerline.

Let’s first look at a conventional setup. Imagine you’re sitting in the normal sweet spot, speakers facing approximately straight ahead, maybe toed in a little. Perhaps the speakers are ten feet apart. Arrival time and intensity are the same from both speakers, so imaging is good.

Next shift your listening position five feet to one side. The image shifts all the way over to the near speaker. This is because not only are you now much closer to the near speaker, you are also on-axis of the near speaker (or very close to it) and very far off-axis of the far speaker. So both localization mechanisms favor the near speaker. In fact if you only shift partway over, the center vocalist’s image usually shifts farther than you do.

Now let’s try something totally different: First, we start out with speakers that have a very specific radiation pattern: The radiation pattern is 90 degrees wide in the horizontal plane (-6 dB @ 45 degrees off-axis to either side), and this pattern is pretty much constant over as much of the spectrum as is practical (down to 700 Hz would be nice, but down to 1.4 kHz still works well).

Second, we toe those speakers in severely, like maybe 45 degrees, such that their axes actually criss-cross in front of the central sweet spot. Yes it looks weird, but stick with me.

In the central sweet spot, arrival time and intensity is the same for both speakers. But now let’s move over five feet to one side...

Now we are sitting directly in front of the near speaker, so it "wins" arrival time. But we are also very far off-axis of that near speaker. We look over at the far speaker, and by golly we’re just about right smack on-axis of the far speaker! And so the far speaker "wins" intensity! These two psychoacoustic localization mechanisms balance out somewhat, so we still end up with a decent spread of the instruments. Depending on the recording and a few other details, the center vocalist may still be fairly close to the center. Now the soundstaging isn’t going to be as good as it is up and down the centerline, but it’s going to be way better than what we had with a conventional set-up.

The KEY to this working well is, the output of that near speaker must fall off quickly and smoothly as we move off-axis, at least in the mids and highs were we get most of our imaging cues. This crossfiring setup doesn’t work very well with conventional speakers because they don’t have the right kind of radiation pattern - the near speaker’s off-axis response is still too loud.

At audio shows I try to set one chair up against a side wall, actually to the outside its nearest speaker. When the room is full and someone is forced to take that seat, I give them a couple of minutes there and then ask how it sounds. I’ve never had anyone be anything other than pleasantly surprised at how good it sounds even from such an extreme off-centerline location.

Another advantage of this configuration is, considerably more uniform tonal balance throughout the listening area. In particular, the cross-firing configuration results in a more uniform distribution of the highs, so nobody gets cheated in that regard.

One more advantage of this configuration is, it minimizes detrimental early same-side-wall reflections. The first significant sidewall reflection for the left speaker is the long across-the-room bounce off the right side wall, and vice-versa.

There is a slight trade-off: Best imaging for one person in the sweet spot is arguably a bit better with the speakers aimed right straight at the listener’s ears, or maybe aimed at a point a foot or two behind the listener’s head. Some taming of the top-end energy may be needed, as now you are directly on-axis of both tweeters, whereas with the criss-cross setup you are never directly on-axis of both tweeters. The tonal balance will also be less consistent throughout the room when the setup is optimized for a single listener.

So if your priority is "10 foot wide good sound", imo it can be done with the right kind of speakers in the right kind of configuration. Examples of speakers that can do it are the JBL M2 and 4367, the PBN M2!5, anything by Earl Geddes, most models by PiSpeakers, and most of my stuff. I’m sure there are others that don’t come to mind offhand.

Duke

@fleschler  wrote:  "I couldn't imagine the speakers at a 45/45 angle towards me."

I can understand that that's just too much visual weirdness for some people.

The first photo in this show report is taken from well off to one side of the centerline.  As you will see, in that location you are well off-axis of the near speaker but nearly on-axis of the far speaker. 

https://parttimeaudiophile.com/2015/10/22/rmaf-2015-audiokinesis-violates-space-and-time/

Unfortunately the write-up doesn't mention the sweet spot width.  But it does talk about soundstage size, in case your first instinct is that the soundstage would be compressed. 

"Among the best disappearing speakers are omnidirectional speakers."

I agree.  With omnis, the near speaker still "wins" both arrival time and intensity, though it doesn't win the latter by as large a margin as with conventional speaker.  So I think there is an argument for the approach I described even compared with omnis because with my approach the far speaker is the one that "wins" intensity. 

But you'd have to get past the weirdness, and that may be asking too much. 

Duke

Back on August 31st I posted that "getting the reverberant field right" matters if rich and natural-sounding timbre is a high priority.

I think there was a fair amount of skepticism, with @prof expressing it well back on September 8th: "I’ve never heard more room sound contribute to more accurate timbre."

I’d like to offer a youtube video of a presentation by acoustician David Griesinger. David investigates concert hall acoustics, and he has determined which reflection are beneficial and which are detrimental based on timing. He is going to demonstrate this by playing four clips that include and/or exclude early and late reflections.

First, he will play the direct sound only. This clip is time-gated to exclude all reflections, so it sounds thin because the longer wavelengths are also excluded. The singer’s voice sounds "proximate" (up close) because of the lack of reflections.

Next, he will play the direct sound plus the first reflections. The timbre will be a bit warmer because longer wavelengths are included, but the clarity will be significantly degraded.

The third clip is by far the most interesting: Direct sound MINUS the early reflections but INCLUDING the later ones. Now we have clarity along with our elusive friend, Rich Timbre!

The final clip includes them all: Direct sound + first reflections + later reflections. Timbre and clarity are both degraded relative to the third clip, but timbre is still better than the first and second clips.

Here’s the video, start at 13:19 and go to about 15:02, headphones or earbuds recommended:

https://www.youtube.com/watch?v=84epTR2fyTY

Just for fun, go back to 14:20 and listen to that third clip again. How about THAT timbre?? Imo THAT is the target!

I believe the psychoacoustic principles demonstrated in David Griesinger's clips have validity that extends beyond the concert hall and into our home listening rooms.

As these clips show, EARLY reflections are detrimental, but LATER reflections can be quite beneficial, enriching timbre with no degradation of clarity. So "more room sound" CAN result in "more accurate timbre" IF it is done right.

Duke