Why evaluating the sound of a single speaker is essential

[Heinrich]

So bad behaviour of the speaker, such as resonances, non-linear distortion, anomalies in directivity - they get caught in a single speaker test.

Yep, agreed. The mono test in controlled environment seems to be a great tool during development. In my setup it doesn‘t work, though. The sound simply doesn‘t feel right. Speakers, by the way, are top rank and reportedly flawless, are equalized to ruler flat in direct field, and are supplemented by ‚perfect subs‘, for each its own stereo. The bigger listening room shows an RT30 of 500ms, but it is pretty flat over frequency. Single speaker mono sounds thin, scratchy even, unsatisfying, meh. Early reflections - may be. Or bass is missing despite an extension to nearly 20 Hz. Stereo is fine.

 

[goat76]

Why does the full-range system have to be compromised?

No, I said that translation is never about compromising the sound for a better full-range sound system.

I would have thought recordings mixed/mastered on a transparent full range system and ( ideally) reproduced on a transparent full range system would be the idea.
Keith

Yes, a mix should always sound the very best on a transparent full-range system. A mix with the best possible translation in the midrange will also make sure it sounds even better on a transparent full-range system.

I give you one example out of many...
A kick drum in a music mix can sound "good enough" in a full-range system, but can at the same time be almost totally absent when the mix is played in a bass-limited system. This often means that the fundamental frequencies are there for the kick drum, but the overtones and the midrange energy may be lacking. You could say that flaws in the mix "got away" with an unnatural balance in the midrange frequency area, even though the kick drum sounded pretty decent while only monitoring the mix in the full-range system.
When addressing this problem in the mix and by making sure the full range sound of the kick drum is naturally heard, which mean it should also sound well-balanced higher up in the midrange area, the mix will better translate to a bass-limited system, but not only that, it will also sound even better when played in a full-range system.

 

[amirm]

When addressing this problem in the mix and by making sure the full range sound of the kick drum is naturally heard, which mean it should also sound well-balanced higher up in the midrange area, the mix will better translate to a bass-limited system, but not only that, it will also sound even better when played in a full-range system.

A bass heavy track will sound bright if you take the lower end away. There is no fix for this other than compromising the full range system by having less treble. If you boost the upper bass to compensate for the bass shy system, then the full range system will sound tubby. All of this is easily proven by doing some simple EQ. What I do in every review.

The only real fix is to let the playback system be in charge. Produce full range music and let the player perceptually correct for lack of full range response to listener preference. This is trivially done instead of relying on "translation magic" which can't possibly work without a ton of compromises.

The world that was a mono speaker in a TV or crappy system in car audio is no more.

 

[respice finem]

Having too loud bass can also be problematic, human hearing is a bit "strange" sometimes:

Auditory masking - Wikipedia

en.wikipedia.org

 

[Newman]

The mono test in controlled environment seems to be a great tool during development. In my setup it doesn‘t work, though.

Hang on a minute.

The sound simply doesn‘t feel right. Speakers, by the way, are top rank and reportedly flawless, are equalized to ruler flat in direct field, and are supplemented by ‚perfect subs‘, for each its own stereo. The bigger listening room shows an RT30 of 500ms, but it is pretty flat over frequency. Single speaker mono sounds thin, scratchy even, unsatisfying, meh. Early reflections - may be. Or bass is missing despite an extension to nearly 20 Hz. Stereo is fine.

Irrelevant. The main point is, use it for comparisons between speakers, not for comparisons between mono and stereo preference.

 

[goat76]

A bass heavy track will sound bright if you take the lower end away. There is no fix for this other than compromising the full range system by having less treble. If you boost the upper bass to compensate for the bass shy system, then the full range system will sound tubby. All of this is easily proven by doing some simple EQ. What I do in every review.

The only real fix is to let the playback system be in charge. Produce full range music and let the player perceptually correct for lack of full range response to listener preference. This is trivially done instead of relying on "translation magic" which can't possibly work without a ton of compromises.

The world that was a mono speaker in a TV or crappy system in car audio is no more.

We are talking past each other if you think I’m talking about overall tonal balance.

The overall tonal balance should always be well-ballanced no matter what system the content is played on. The question I am answering is about mixing translation, and that is mostly about balancing and EQing the sound objects within a mix. That's what mixing engineers are talking about when it comes to translation.

 

[Mark617]

We are talking past each other if you think I’m talking about overall tonal balance.

The overall tonal balance should always be well-ballanced no matter what system the content is played on. The question I am answering is about mixing translation, and that is mostly about balancing and EQing the sound objects within a mix. That's what mixing engineers are talking about when it comes to translation.

Misaligned audio luminaries?

 

[Duke]

You didn't misunderstand, but you need to go back to the original paper to see what drove the preferences when the room dominated. I show these data in the 4th edition and discuss them, showing that the preferred room/loudspeaker combinations had the most bass and the least preferred had the least bass. The bass in all cases was very rough because there had been no effort to tame the room resonances. As I keep saying, bass is important, and smooth bass is better than resonant bass..

Thank you for clarifying; I didn't realize that the perceived sound quality difference between the rooms correlated with which room had the most bass. I was mistaken in thinking that significant differences existed further up the spectrum.

At high-end audio shows we have two categories of rooms: "Small rooms" which are the normal hotel rooms, and "big rooms" which are either large suites or conference rooms or something like that, and ime there are differences in sound quality and spatial quality between the small rooms and the big rooms which extend beyond the bass region. Of course this may well be largely because usually (though not always!) bigger and more expensive speakers are being shown in the big rooms.

Imo it is remarkable when the sound quality and/or spatial quality in a small room is competitive with the big rooms.

This observation seems to hold up over the scores of audio shows I've attended, but of course it is a SIGHTED observation and therefore inevitably subject to bias. So as a reality check, I hope you don't mind me asking:

In your opinion, is there a sound quality and/or spatial quality advantage above the bass region that comes from a big room's longer reflection path lengths?

 

[Floyd Toole]

In your opinion, is there a sound quality and/or spatial quality advantage above the bass region that comes from a big room's longer reflection path lengths?

I wish I had a simple answer to that good question. Chapter 7 in the 3rd edition discusses the audibility of early lateral and vertical reflections for several different sounds - and it makes a difference. I will try and find some rational comments when I finish examining the page proofs of the 4th edition, which discusses the topic as well, and then compile a useful index - a difficult job. Business first

Off the top of my head, I would estimate that there is little experimental evidence to support advantages to either large or small rooms above the transition frequency. But having lived with both, I favour a large room in part to be able to get farther from the loudspeakers (the musical instruments). That said, we need to get specific about the dimensions at which there is a transition from small to large. Good luck with that.

 

[MattHooper]

But having lived with both, I favour a large room in part to be able to get farther from the loudspeakers (the musical instruments).

When you get back and have more time to answer, I’d be interested in you expanding a bit on that remark.

I’m wondering, if getting further from the loudspeakers is a personal preference of yours or whether it’s based on some larger principal for better sound.

I have generally favoured being somewhat close to my loudspeakers, between 8 and 6/12 feet with 7 feet or 7 1/2 feet being my sweet spot. I like the sense of immersion.

Since I’ve always listened in the same room, which is 15’ x 13’ that certainly could’ve influenced my preferences over the years.
In any case, even when I’m listening to systems in larger rooms I still tend to prefer getting around that distance from the loudspeakers, at least as close as 8 feet.
(unless I’m listening to a loud speaker design that doesn’t sound very good unless I get further away).

 

[thewas]

I have generally favoured being somewhat close to my loudspeakers, between 8 and 6/12 feet with 7 feet or 7 1/2 feet being my sweet spot. I like the sense of immersion.

I think your preference (mine and few audiophile friends of mine is also similar) of being more in the acoustic nearfield where reflected sound doesn't dominate over the direct one which for typical relatively wide radiating Hifi loudspeakers is around 2 meters, has also to do that we listen to mainly stereo setups, while Dr. Toole listens mainly to multichannel where the immersion is created more through the different directions.

 

[Duke]

I wish I had a simple answer to that good question. Chapter 7 in the 3rd edition discusses the audibility of early lateral and vertical reflections for several different sounds - and it makes a difference. I will try and find some rational comments when I finish examining the page proofs of the 4th edition, which discusses the topic as well, and then compile a useful index - a difficult job. Business first

Off the top of my head, I would estimate that there is little experimental evidence to support advantages to either large or small rooms above the transition frequency. But having lived with both, I favour a large room in part to be able to get farther from the loudspeakers (the musical instruments). That said, we need to get specific about the dimensions at which there is a transition from small to large. Good luck with that.

Thank you for taking the time!

Someone smarter than me will have be the one to draw the line between "small rooms" and "large rooms". At audio shows the distinction is fairly easy to make, as the "large" rooms are typically at least twice the size of the "small" ones (and usually the difference is greater than that). On the other hand the "real world" of high-end home audio includes a lot of in-between-sized rooms as well as many considerably smaller ones.

I remember reading somewhere, perhaps in an AES paper (though for some reason I vaguely recollect it being European), either a proposal for a 15 milliseconds Initial Time-Delay Gap as a target for studio control rooms, or maybe the paper cited 15 milliseconds as an already-established target (I should have paid closer attention). I've seen 10 milliseconds on the low side, and 20 milliseconds on the high (and presumably more desirable) side, mentioned in the context of Reflection-Free Zone control rooms.

If there is an Initial Time-Delay Gap which results in a spatially or sound-quality-wise preferred presentation, perhaps that might figure into making a distinction between "small rooms" and "large rooms" in a home audio context. However I suspect that if the experimental evidence does not yet exist, then it's unlikely to ever exist, as funding has probably shifted to other priorities.

 

[MattHooper]

I think your preference (mine and few audiophile friends of mine is also similar) of being more in the acoustic nearfield where reflected sound doesn't dominate over the direct one which for typical relatively wide radiating Hifi loudspeakers is around 2 meters, has also to do that we listen to mainly stereo setups, while Dr. Toole listens mainly to multichannel where the immersion is created more through the different directions.

Yes, I certainly considered that.

As I mentioned in my own experience since I have both my two channel speakers and my surround system in the same room I can directly compare them or mix and match.

I certainly get the sense of surround immersion from the surround sound. But I get a different sense of immersion from the two channel speakers pulled close to me for stereo listening.

I sit about 9’7” from my centre channel and about 11 feet to the left right speakers.

My two channel floor speakers are pulled out so they are about 7 feet from the listening position and spread 8 feet.

Even though I get a sense of immersion when I play my surround system, the sound still starts further away from me in terms of all the main information at the front of the stage.

It’s very easy for me to take the leads from my left/right speakers and put them into my two channel floor standing speakers to have them take over L/R duty. When I do that I find the sound even more immersive and satisfying.

So in my case, it seems to me even if I were going strictly surround for music listening, my best case scenario would still have at least the L/R speakers at a similar distance to what I’m using for stereo.

(as I mentioned 7 feet is about my sweet spot - I’m not a fan of truly near field, listening… for me if I get too near field, it starts to lose its “live” feeling and becomes more headphone like)

 

[LSPhil]

Yep, agreed. The mono test in controlled environment seems to be a great tool during development. In my setup it doesn‘t work, though. The sound simply doesn‘t feel right. Speakers, by the way, are top rank and reportedly flawless, are equalized to ruler flat in direct field, and are supplemented by ‚perfect subs‘, for each its own stereo. The bigger listening room shows an RT30 of 500ms, but it is pretty flat over frequency. Single speaker mono sounds thin, scratchy even, unsatisfying, meh. Early reflections - may be. Or bass is missing despite an extension to nearly 20 Hz. Stereo is fine.

I’ve been away on business for a month, and I took a small setup with me: a 2-way speaker using the SB12MNRX midwoofer and a Tymphany D27TG35-06 tweeter. I crossed them at 1.5 kHz with an acoustic Butterworth slope of 42 dB/oct. On the tweeter I applied a Linkwitz correction to one of the 12 dB filters, and everything measured fine.

At first, I only connected it to the left channel, but the sound felt thin compared to the rest of the spectrum. I thought to myself: if I’m going to listen in mono anyway, I might as well feed both channels in. The MiniDSP has two inputs (1 = left, 2 = right), so I summed them – and suddenly everything fell into place. The music just worked: no harshness, nothing fatiguing, just hours of enjoyable listening.

What this tells me is that we really must listen in mono when evaluating a single speaker. The same thing happens when listening directly at a PC – mono is created in the centre by summation, while stereo gives almost a surround effect with satellites plus a mono centre.

Back when I was experimenting with a DCX2496, I ran a single mono loudspeaker and swept the crossover frequency from 200 Hz up to 3 kHz, using a Seas L12 and a Seas MU10. What I found was that the best results didn’t appear until the crossover reached about 1 kHz. In fact, almost all speakers that cross around 1 kHz have that mysterious quality – unless you’re using a really good dedicated midrange driver for the 200 Hz – 1 kHz band.

So my loose conclusion is that it’s important to have the same driver covering that whole 200 Hz to 1 kHz region.

Here I’ve given two examples from my own experience of listening to loudspeakers in mono.

 

[Matt_Holland]

So my loose conclusion is that it’s important to have the same driver covering that whole 200 Hz to 1 kHz region.

You will approve of the direction the Ascilab C6 tower development is heading in.

 

[Heinrich]

... The MiniDSP has two inputs (1 = left, 2 = right), so I summed them – and suddenly everything fell into place. The music just worked: no harshness, nothing fatiguing, just hours of enjoyable listening.

What this tells me is that we really must listen in mono when evaluating a single speaker. The same thing happens when listening directly at a PC – mono is created in the centre by summation, ...

What you say is, that listening in mono is listening to both channels summed. That is even farther away from the recoding engineer's intentions, when it come to strange interference in stereo setup, or the mitigation of it.

My take is, that it may be beneficial to listen 'mono' either way in regard to isolated properties of the speaker part of the set-up. As a real-world customer I'm not particularly interested in such isolated criteria, though. Any evaluation should at least in one test focus on the devices' intended use case--me thinks that's obvious. Maybe some other tests may contribute to the whole story. I doubt that too many adepts at sound perfection find the time and appreciate the effort to perform many tests with reliable (objective) reporting on each of them Hence, one may get away with the crucial single test in stereo (or whatever the set-up will be eventually).

 

[Duke]

Back when I was experimenting with a DCX2496, I ran a single mono loudspeaker and swept the crossover frequency from 200 Hz up to 3 kHz, using a Seas L12 and a Seas MU10. What I found was that the best results didn’t appear until the crossover reached about 1 kHz. In fact, almost all speakers that cross around 1 kHz have that mysterious quality – unless you’re using a really good dedicated midrange driver for the 200 Hz – 1 kHz band.

So my loose conclusion is that it’s important to have the same driver covering that whole 200 Hz to 1 kHz region.

Interesting.

If you look at a set of equal loudness curves, you'll see that in the 70-100 phon region they have a gentle bump around 1 kHz. This indicates a corresponding dip in the ear's sensitivity in that region, imo implying that 1 kHz ballpark might be a good place for a crossover from the standpoint of psychoacoustics:

By the same reasoning, the 3-4 kHz region (and maybe somewhat to either side) looks like an especially challenging place for a crossover, as this is where the ear is most sensitive and likely to be the least forgiving.

And yes, ime this is the sort of thing that can be more reliably evaluated by listening to a single speaker than by listening to a stereo pair.

 

[Salt]

Interesting.

If you look at a set of equal loudness curves, you'll see that in the 70-100 phon region they have a gentle bump around 1 kHz. This indicates a corresponding dip in the ear's sensitivity in that region, imo implying that 1 kHz ballpark might be a good place for a crossover from the standpoint of psychoacoustics:

View attachment 473632

By the same reasoning, the 3-4 kHz region (and maybe somewhat to either side) looks like an especially challenging place for a crossover, as this is where the ear is most sensitive and likely to be the least forgiving.

And yes, ime this is the sort of thing that can be more reliably evaluated by listening to a single speaker than by listening to a stereo pair.

This was addressed by 3 way speakers with mids from 200 or 300 Hz up to 3 or 4 kHz.
But somewhere on the road to High-end we ended at 2.5 kHz midwoofer to tweeter in mainstream, and that over decades.
Ascilab et. al. are now enforcing a new perspective, successfuly, but my recent project will be something completely different.

 

[Floyd Toole]

Interesting.

If you look at a set of equal loudness curves, you'll see that in the 70-100 phon region they have a gentle bump around 1 kHz. This indicates a corresponding dip in the ear's sensitivity in that region, imo implying that 1 kHz ballpark might be a good place for a crossover from the standpoint of psychoacoustics:

View attachment 473632

By the same reasoning, the 3-4 kHz region (and maybe somewhat to either side) looks like an especially challenging place for a crossover, as this is where the ear is most sensitive and likely to be the least forgiving.

And yes, ime this is the sort of thing that can be more reliably evaluated by listening to a single speaker than by listening to a stereo pair.

Which equal loudness curves are these? They look like the last generation ones based on Robinson and Dadson, not the latest ones. Equal-loudness contours are not engineering data. Pure tones are not music. Anechoic chambers are not listening rooms.

 

[Duke]

Which equal loudness curves are these? They look like the last generation ones based on Robinson and Dadson, not the latest ones.

I don't remember where it comes from; I saved it years ago, and just now was unable to find it online again.

I did not realize how different the more recent curves are!

Equal-loudness contours are not engineering data. Pure tones are not music. Anechoic chambers are not listening rooms.
View attachment 473670

Now the bumpage seems to be centered on about 1.5 kHz, which is arguably good new for speaker designers: Seems to me it's usually easier to cross over at 1.5 kHz than down around 1 kHz.