omnidirectional loudspeakers = best design available

[Wesayso]

Your sarcasm was neither appreciated nor necessary, and I see Dr Toole has done me — and more importantly, the thread — a huge favour in his posts above.

Cheers

Indeed he did and I'm thankful for that. It might mean that your task is less needed though. Meaning you don't need to be his spokesperson.
But old habits and all that, I see you're back at making your own interpretation already. That's the part I could do without.

 

[Newman]

@Wesayso it was actually a dialog with Dr Toole, how can you possibly object to my checking with Dr Toole that I am summarising my understanding of his conclusions correctly?

Weird. But hey it’s the internet.

I guess you would also prefer I never made this comment about interpreting one of Dr Toole’s datasets? Maybe you need to get off my back a little?

 

[youngho]

@Floyd Toole if I may, can I make the question a bit more specific? I've asked Dr. Geddes and I always wanted to ask you the same thing.
The question raised in this thread was, in a very broad and/or general way, does the type of music the listener prefers have any relation to the preferred DI?
Like Dr. Geddes mentioned when I asked him earlier about 'early reflections' and the difference in opinion between what he prefers and your point of view on it (see quotes a few posts back):

@Wesayso I think it's the recording and mixing techniques, rather than the type of music per se (since classical recording techniques can run the gamut, rather simplified here but may be helpful:http://www.playclassics.com/trtsound), that are relevant. You might make some inferences from what Toole says here: https://gearspace.com/board/showpost.php?p=15187387&postcount=61

As @david moran pointed out earlier, much of what is discussed in ASR as narrow or wide DI (along with beamwidth) occurs fairly high in the frequency spectrum when it comes to musical content, as well as the human voice, though I hadn't really considered how the transition to omni radiation for most conventional speakers may occur relatively close to the transition zone for increasingly modal behavior for many domestic listening rooms. I also have an informal theory that there may be a minority of listeners who may be particularly sensitive to something like timbre when it comes to a critical frequency range from a few hundred to several thousand hertz, so perhaps they may prefer a relatively flatter DI (as in many that emerged from Harman, like the https://www.audiosciencereview.com/forum/index.php?threads/revel-f228be-review-speaker.23659/) versus a more steadily rising one (common in Kef and Genelec speakers, like https://www.audiosciencereview.com/...s/genelec-8361a-review-powered-monitor.28039/).

It occurred to me that someone with access with an acoustic virtual reality lab like the two in Finland (https://www.genelec.com/-/bang-olufsen-chooses-genelec-for-its-virtual-reality-laboratory and https://www.aalto.fi/en/aalto-acoustics-lab/aalto-acoustics-lab-facilities) could potentially explore a number of interesting questions when it comes to speaker behavior and room acoustics in terms of rapid comparisons "virtually" impossible in the real world but really possible in the virtual one.

Young-Ho

 

[Floyd Toole]

No my question isn't answered at all by appeal to music sales. I'm talking about sound quality, not music quality. People can appreciate music on all sorts of devices, from old transistor radios, crappy turntables, laptops or name the "worst measuring" speaker brand you want. Whatever people like musically doesn't help tell us anything about sound quality.

One way to think about my question is to ask yourself: If you wanted to demonstrate your hi-fi system to a non-audiophile, that is show what it can do as opposed to much poorer quality systems most people might be used to, what demo tracks would you select and why?

Personally, I've used various demo tracks over the years. I can put on some recordings - e.g. vividly recorded vocals with light acoustic accompaniment (e.g. some Astrud Gilberto tracks, virtually everyone likes the music) and every time...every single time...the listener is blown away. All rate the sound quality they are hearing as very high. Why? Always the same reaction "It Sounds So Real!!!" Most people never even have an inkling of what a high end system can do in sonic terms, or even think music can "sound realistic" so they tend to be shocked to encounter such sound. (Of course I'm not talking perfectly indistinguishable from live - as Dr. Toole says, stereo is compromised. But certainly "much CLOSER to the real thing" than most people have heard).

So, if you were to show a non-audiophile what your system is capable of...why you have put so much time, attention (and possibly money relative to a non-audiophile) in to it...what type of recordings would you use? Presumably you wouldn't use a horrendous, thin, scratchy recording right? So just think about the general attributes that you think will tend to bring forth a response "that sounds really good." What would those be?

ETA: Apparently Harman has a list of recommended tracks:

Harman Art of Listening Playlist

Harman Art of Listening Playlist · Playlist · 24 songs · 141 likes

open.spotify.com

So it would be relevant to ask: is there a general trend in there in terms of why they were chosen, such that they would reveal what people will judge as "good sound?" (I've given my own answers on this before, but I'm interested in Dr. Toole's view, and certainly in yours too Newman).

It's a bit of a philosophical chicken-or-the-egg question. What Makes For Good Sound Quality? You can't say "speakers that measure well" because they won't sound "good" to people if the recordings themselves sound "bad." But then, it can't just be the recording itself: if you play a "good recording" through a "bad speaker" then you'll get "bad sound." So then what is the measure of a "good recording?" For sound reproduction quality, neither appeal to the recording itself, nor the speaker itself, can tell us what Good Sound is. Which implies we have some sort of notion of "good sound when we hear it" criteria. (And at least one viable criteria - maybe not the only - would be "sounds like the real thing." I'm sure there are other touchstones too).

I think you will find answers to your questions in the 3rd edition of my book. The beginning of Chapter 18 describes how I stumbled into a lifetime of psychoacoustic research by asking essentially the same question you did: "what is good sound, and is there agreement on the objective?" I had some advantages over most people. I had access to an excellent anechoic chamber in which to do measurements, although I did not know how to interpret the data beyond the most naive notions. I also arbitrarily chose to compare four loudspeakers at a time in an A, B, C, D fashion. My experience with my PhD project listening tests on sound localization taught me that loudness had to be equalized, presentations had to be randomized, and the tests had to be "blind". Having multiple loudspeakers in the comparison made the listener's task much easier, because very quickly timbral signatures of program material (constant) could be separated from timbral contributions of the loudspeakers (variable). This method has continued to this day.

Recordings are enormously variable, so what makes a good one? Here we must distinguish between a good "demo" recording and a revealing "test" recording. Years of listening to "audiophile" recordings tell me that most are in the "demo" category. Isolated clear voices with simple accompaniments are always included. From our observations over the years voices and solo instruments are not very revealing of timbral problems. We learned much later that bass quantity and quality account for about 30% of an overall subjective sound quality rating, so recordings with good bass extension are useful, and dense orchestration with some reverb assist in revealing coloration due to resonances. Much to our dismay, the classical repertoire did not distinguish itself, but pop/rock with high production values did.

Listeners could not possibly know what such studio recordings should sound like, so how could they make reliable, repeatable ratings of sound quality. It was not because they were recognizing excellence. It seems most probable that they were responding to unnatural "foreign" contributions to the recordings, by far the most common one being acoustical and mechanical resonances in the loudspeakers. By this reckoning, the "best" sound was in fact the "least bad" sound. "Perfection" is revealed when audible colorations are attenuated. This was confirmed in measurements, with the currently standardized "spinorama" being a presentation capable of describing enough of the sound arriving at a listening position to be able to calculate a trustworthy prediction of subjective ratings. Timbral neutrality, it seems, is a necessary starting point for loudspeaker design. Beyond that spectral balance matters, and this is the "tone control" region of equalization for variations in recordings and loudspeaker/room interactions.

Now, what about directivity. The notion that omnidirectIonality is somehow "ideal" awaits proof. As described in my book, I have enjoyed the sound from both omnis and conventional forward-firing loudspeakers. Recordings are not made with omni monitors, so their use is that of an embellishment - a "sound effect" that in some situations and with some program compensates for limitations of stereo. With multichannel recordings or tasteful multichannel upmixing of stereo material omnis cease to be advantageous. They do not, as has been claimed, have any "natural" sound connection. In fact all sound sources that matter to music lovers exhibit directivity similar to forward firing loudspeakers - see the illustration.

 

[fineMen]

We learned much later that bass quantity and quality account for about 30% of an overall subjective sound quality rating, so recordings with good bass extension are useful, ...

I first argued that better quality in bass coincides with overall better speakers. The necessarily bigger enclosure sets the cost so that it is economical to invest in parallel in better drivers and all. But for the time being I'm quite happy with a sub-shoebox-sized speaker box that is equalised clean to a f-3 of just 30Hz. A virtually unlimited bandwith is preferable, if the rest is likewise o/k ;-)

... and dense orchestration with some reverb assist in revealing coloration due to resonances. Much to our dismay, the classical repertoire did not distinguish itself, but pop/rock with high production values did.

The infamous 'classic' is designed to give a full sound from many instruments which shall not appear as single entities. It is not about the instrument but that hefty wall of sound it contributes to. Additionally the focus is not the timbre, but the harmonic development. For instance I never encountered a playback that ruined Anton Webern's (cont. of Schoenberg) songs.

The little speaker mentioned above is a fully worked out 3-way. The midrange is devoid of any modulation from the bass. And again, smaller speakers regularly are both, bass restricted and, as two-ways, loaded with intermodulation.

Listeners could not possibly know what such studio recordings should sound like, so how could they make reliable, repeatable ratings of sound quality. It was not because they were recognizing excellence. It seems most probable that they were responding to unnatural "foreign" contributions to the recordings, by far the most common one being acoustical and mechanical resonances in the loudspeakers.

But music is about resonances. Interesting music has more of them, and especially unheard ones. Hence I would substitute: resonances / colorations that are the same with every program. My unspecific first guess. This opens another question. Could it be so that people got used to objectively bad speakers, that they learned not only to accept the deficiencies, but expect them to be present as a quality marker? If not present as some funky spice to the boring dish, no drama, no good speaker.

Now, what about directivity.

Let me tell you, that I really needed your book. It was the best thing I read about stereo ever. Thank you!

I've got a question though. The direct sound is about the directional cues, the localisation of the virtual sound source. The overall sound, reverberation included, is about the timbre. A not too severe imbalance in the direct sound would only harm the stereo 'imaging', if the balance of the overall sound is maintained.

Would You confirm?

Why not separate the tasks of (a) throwing directional cues at the listener and (b) filling the room with well balanced sound?

On topic?

.

 

[Floyd Toole]

On topic, indeed.

Musical timbre is most certainly defined by resonances - the "tune" and the tonal identifiers of voices and instruments. That is why we don't like to hear that complex structure modified by additional resonances in loudspeakers and rooms (at low frequencies). If they are sufficiently audible they become monotonous annoyances.

Separating tasks. It turns out that the direct sound does more than define direction. It becomes the initiation of the precedence effect, in which later arrivals are audible but do not substantially modify the localization cue. It turns out also that the direct sound is a major factor in determining sound quality ( see part 3 of the Sean Olive experiments described in Section 5.7.3, p. 140 in the 3rd edition.). It seems to be a reference against which later arrivals of the same sound are compared.

From the earliest experiments I did, the principal identifier of a "neutral" loudspeaker was its on-axis - direct sound - performance. If the off-axis performance was also smoothly non-resonant the ratings were even higher. The following figure is Figure 13.1 from the 3rd edition. It shows 30 years of progress in engineering resonances out of loudspeakers. Please do not interpret the "idealized room curve" as a "target" for room EQ. It isn't. It is the anticipated room curve from the measured product, in this case well designed ones. It is not an EQ target for flawed loudspeakers - that cannot be reliable. The explanation is in the book.

 

[Thomas_A]

It seems to me that too many appear to equalize their room response to a target. As Floyd say there is no ”target”, just a result: i.e. a response of a good measuring speaker in an average room.

 

[gnarly]

It seems to me that too many appear to equalize their room response to a target. As Floyd say there is no ”target”, just a result: i.e. a response of a good measuring speaker in an average room.

I totally get the idea of a natural preference for a downward sloping frequency response (magitude) target curve, for a "best-fit" to any randomly chosen recording.
I have that preference too.

But for me, the idea of having to live with a fixed, non-adjustable target curve.....well, no dice, no way.
Recordings simply vary too much in their tonal balance, at medium playback levels..
Our various playback levels change perceived tonal balance further.

I'm really saying I think good ole, real-time, tone controls are a must (for me anyway)

 

[Blumlein 88]

I totally get the idea of a natural preference for a downward sloping frequency response (magitude) target curve, for a "best-fit" to any randomly chosen recording.
I have that preference too.

But for me, the idea of having to live with a fixed, non-adjustable target curve.....well, no dice, no way.
Recordings simply vary too much in their tonal balance, at medium playback levels..
Our various playback levels change perceived tonal balance further.

I'm really saying I think good ole, real-time, tone controls are a must (for me anyway)

One thing often forgotten about that target curve. It is a result of how measurements are done in a room gated. Anenchoically a speaker that measured dead flat will measure with that target curve slope or something like it in room. So it really isn't a general roll off of the signal being fed into it.

I do agree recording balances vary and some control for that can be very useful. I've wondered if having one of the old pro simple EQ units might work well. One that has two or three bands to play with just for general program material adjustment. When I used to run a Tact Room Correction preamp, it had 9 presets. I had one that was flat transparency and several that tilted up or down. You could switch instantly from the remote so adjusting for various recordings issues was pretty simple. I also think for simplicity the old Quad idea of tilt rather than Bass and Treble is what made the most sense most of the time.

 

[dlaloum]

I totally get the idea of a natural preference for a downward sloping frequency response (magitude) target curve, for a "best-fit" to any randomly chosen recording.
I have that preference too.

But for me, the idea of having to live with a fixed, non-adjustable target curve.....well, no dice, no way.
Recordings simply vary too much in their tonal balance, at medium playback levels..
Our various playback levels change perceived tonal balance further.

I'm really saying I think good ole, real-time, tone controls are a must (for me anyway)

I think the best tone control I have ever experienced was the Quad "Tilt" control on the 34 and 44 preamps...

 

[Floyd Toole]

I totally get the idea of a natural preference for a downward sloping frequency response (magitude) target curve, for a "best-fit" to any randomly chosen recording.
I have that preference too.

But for me, the idea of having to live with a fixed, non-adjustable target curve.....well, no dice, no way.
Recordings simply vary too much in their tonal balance, at medium playback levels..
Our various playback levels change perceived tonal balance further.

I'm really saying I think good ole, real-time, tone controls are a must (for me anyway)

You seem not to have noticed that a downward tilted "steady-state" room curve results from loudspeakers that radiate absolutely horizontally flat direct sound. It is the direct sound that is the more definitive factor in sound quality evaluations. That is the main reason why steady-state room curves are not reliable indicators of sound quality. One needs anechoic data on the loudspeaker. But try to convince the marketing folks for these algorithms that they are not the perfect cure, but instead a source of confusion. That said, Room EQ works at low frequencies.

Spectral balance variations in recordings are abundant. I have for decades argued for the liberal use of tone controls - but sadly some under-informed "high end" folks think they are the work of the devil, and they cannot be found in all electronics. They also need to be easily accessible, not buried in layers of software menus.

The "tilt" control in the old Quad preamp was another step in the right direction. It was a crude tilt, being the result of linking the tone controls - bass boost/treble cut, or bass cut/ treble boost. Much later, Lexicon provided at tilt control in it's processors and because it was digitally created it was truly a linear bass-to-treble tilt. Very useful indeed. One needs such controls if paying attention to spectral balance while listening.

 

[mhardy6647]

Great for parties, but critical listening not so much.
Keith

I really, truly thought this comment was directed at brussels sprouts.

Mrs. H and I both quite like brussels sprouts. They have to be fresh, and properly cooked, or they can be pretty nasty.
Fortunately, they grow quite well in northern New England. We've even had a bit of success growing them ourselves, but we usually get them from the fall CSA to which we belong.

She has a stir fry recipe that uses sprouts and white beans -- and garlic. It is dandy. We had it a couple of nights ago for dinner, in fact.
Roasting (with or without other vegetables) is another good thing to do with them.

My parents, conversely, used to sort of cook them to equilibrium, which was pretty awful.

 

[gnarly]

One thing often forgotten about that target curve. It is a result of how measurements are done in a room gated. Anenchoically a speaker that measured dead flat will measure with that target curve slope or something like it in room. So it really isn't a general roll off of the signal being fed into it.

Thank you for that reminder (and Floyd also, who I'll reply to next).
I needed it as I've been struggling a little with a new DIY speaker build.
It's the first speaker I've built that's so large I can't get it outside, to tune as quasi-anechoically as possible, on a spinorama, ungated & far-field (3-4m)
So I've been making near-field measurements of it indoors, and wondering why am I fighting a perceived brightness vs a flat measured response.

Anyway, I'm going to tune this new big guy indoors with a downward sloping target....simply since I can't get it outside to tune it flat..
(Never measure much indoors (what's a gate, Lol!)
I forget the simple stuff sometimes, mainly out of habits in ways of doing things.

I do agree recording balances vary and some control for that can be very useful. I've wondered if having one of the old pro simple EQ units might work well. One that has two or three bands to play with just for general program material adjustment. When I used to run a Tact Room Correction preamp, it had 9 presets. I had one that was flat transparency and several that tilted up or down. You could switch instantly from the remote so adjusting for various recordings issues was pretty simple. I also think for simplicity the old Quad idea of tilt rather than Bass and Treble is what made the most sense most of the time.

On all previous DIYs, once tuned flat outdoors and then brought indoors, I just use tones controls to tune to taste. Mainly due to recording imbalances, but also due to perceived tonal changes with loudness variations.
Totally agree with the methods you describe. I've built ways of implementing both.
Going to picture them, in reply to Floyd now...

 

[gnarly]

You seem not to have noticed that a downward tilted "steady-state" room curve results from loudspeakers that radiate absolutely horizontally flat direct sound. It is the direct sound that is the more definitive factor in sound quality evaluations. That is the main reason why steady-state room curves are not reliable indicators of sound quality. One needs anechoic data on the loudspeaker. But try to convince the marketing folks for these algorithms that they are not the perfect cure, but instead a source of confusion. That said, Room EQ works at low frequencies.

Thank you. If you read my reply to Blumlien 88, you'll see that yes, i was kinda forgetting anechoic vs in-room response.
An example that has helped me realize just how different anechoic vs in-room is, is a video on the sound of balloon pops made by B&C speakers

I've been taught a balloon pop has an impulse response similar to a Dirac pulse, and pragmatically contains all frequencies, and will measure flat response. (Pls correct if wrong)
The idea that the balloon pop's frequency response will measure the same in an anechoic chamber, as in-room, ......is kind mind blowing for me, given the differences in sound.

Spectral balance variations in recordings are abundant. I have for decades argued for the liberal use of tone controls - but sadly some under-informed "high end" folks think they are the work of the devil, and they cannot be found in all electronics. They also need to be easily accessible, not buried in layers of software menus.

The "tilt" control in the old Quad preamp was another step in the right direction. It was a crude tilt, being the result of linking the tone controls - bass boost/treble cut, or bass cut/ treble boost. Much later, Lexicon provided at tilt control in it's processors and because it was digitally created it was truly a linear bass-to-treble tilt. Very useful indeed. One needs such controls if paying attention to spectral balance while listening.

Yessir. Dialing in spectral balance while listening adds so much enjoyment for me. It's like mixing live-sound, only it's just a recording lol.
And I really like the tilt idea.
In an open-architecture processor, I built a real-time, continuously variable tilt control. Used 700Hz as the center of the spectrum, for a see-saw type tilter.
Here's a few transfer functions showing various levels of tilt.


Unfortunately, the real time adjustment of the filters necessary was a little more than the processor could handle smoothly, mainly because i was asking the processor to do a lot of FIR processing as well.
So I switched to a set of sliders that allow tonal changes pretty easily. It uses a very unusual way of adjusting the ranges, each about 2 octaves...but not really worth going into. I love it.


(I've upgraded processors recently, and won't have any more sluggish tilt with coincident FIR use, so maybe will go back to tilt again.)

 

[Mnyb]

Idea ! Would standardisation help us ? Sometimes perfection seems to the enemy of the good ?

“We” ( collectively as humans, thanks to Dr Toole et al ) seems to know roughly what kind of directivity and other characteristics are the most preferred universally, even as this tread demonstrate not exactly with 6 sigma precision what DI index is “the best”.

Maybe the question is wrong and there is no best but a set good enough but slightly different solutions?

If we simply pick one and stick with it ! The producers and musicians use monitors with this” standard characteristics “
And we listen with similar speakers at home ? It may not be the perfect solution, but it’s a solution and imperfections may very well be compensated be the producers making the music ? It’s at least a stick in the mud

 

[Floyd Toole]

I agree with your "stick in the mud", as is illustrated by the last Figure in my book, shown below. The answer has existed for decades.

 

[Mnyb]

I agree with your "stick in the mud", as is illustrated by the last Figure in my book, shown below. The answer has existed for decades.

You got me , i’m Reading your book , but i’m not at the last chapters yet .
Thanks for making your work aviable for the public. You could have had a “normal” academic research career and never bothered with us plebeians it would have been easier, but you choose this instead.

 

[Floyd Toole]

You got me , i’m Reading your book , but i’m not at the last chapters yet .
Thanks for making your work aviable for the public. You could have had a “normal” academic research career and never bothered with us plebeians it would have been easier, but you choose this instead.

Actually, I often suggest that readers start with the last chapter - of course it is a "spoiler" for some things, but it does set the context for the industry and the topic very well, I think. I did not show any "bad" speakers, and believe me there were a lot. Even now, looking at the published spinoramas - and there are some dandies - one wonders who is listening to what when the product was approved. Still some losers out there.

 

[mhardy6647]

Actually, I often suggest that readers start with the last chapter - of course it is a "spoiler" for some things, but it does set the context for the industry and the topic very well, I think. I did not show any "bad" speakers, and believe me there were a lot. Even now, looking at the published spinoramas - and there are some dandies - one wonders who is listening to what when the product was approved. Still some losers out there.

Remind me not to read any reviews you might publish of mystery novels (or movies).

 

[Mnyb]

I agree with your "stick in the mud", as is illustrated by the last Figure in my book, shown below. The answer has existed for decades.

Makes me wonder what the "flagship speaker" in the graph is ? I'll keep wondering , you show professional courtesy by not name calling anything , suppose the designer of it can see the graph in your book and recognize it