Modern Measurement Tools Are Tricking Audiophiles Into Trusting Bad Data, Warns Veteran Speaker Designer

[Suono]

I would bet the overall sound in that room is big and somewhat incoherent.

as happens in every room, even up to the auditoriums where live music is played.

 

[Suono]

To claim that Andrew Jones, one of the world's best speaker designers, doesn't understand measurements, their importance, mathematics, physics, and their application in acoustics, I find it risky. Perhaps we should measure less and listen more.

 

[Curvature]

as happens in every room, even up to the auditoriums where live music is played.

Absolutely not. Flatly wrong.

If you would like a source: https://www.taylorfrancis.com/books...-absorbers-diffusers-peter-antonio-trevor-cox

 

[Suono]

Assolutamente no.

Grazie.
Se un libro fosse bastato, avrebbero risolto il problema. Invece, ci sono stanze che suonano bene e stanze che non si riescono a far suonare bene. È un processo di perfezionamento costante.

 

[olieb]

Looks to be a diffuse mess.

Yes, but that is actually what diffuse means.

In this room you will have a fairly unpredictable early reflection which is not consistent with the spectrum of the direct sound of the speaker.

I do not see why being strong and specular would make a lateral early reflection a treat. In a concert hall it is not a treat to sit close to a reflecting side wall in my experience.
And why do you think that the "reflection" will have an "inconsistent" spectrum? These QRD diffusers can be calculated and the reflection will be diffused over a wide area. So, if sufficient care is taken, the diffuse lateral sound can certainly be "consistent". From the pictures I would estimate that the room is about 3.5m high and ≥6m wide, so the listener has about 2.5-3m distance to either ceiling or walls.
So the diffusors may "have their own polar patterns" but this is true for a wall, furniture or anything else too, but here it probably is taken into account and the distance is large enough to have the QRDs do their work.
There might be other ways to build a decent listening room but without details or better measurements I do not see how to judge the effort. At least these people seem to know what they are doing.

 

[Curvature]

Based on the "Why do some people say Genelecs are too flattering or forgiving" thread, I was about to start a new thread dedicated specifically to getting knowledgeable members' input on the questions of negative DI in the lower bass region and DI-increase slopes throughout the lower mids to lower treble, because I'm always open to the possibility that maybe I'm just crazy.

But I just stumbled across this thread, which I hadn't seen until now, and it appears @Arindal is flogging the same horse here. To be clear - and to be fair to Arindal - I still would be interested to hear what folks think specifically about Arindal's claim that speakers whose DI increases by more than 2-3dB over a given range* - which includes the vast majority of standmounts from Genelec, Neumann, KEF, Revel, and many others BTW - are both objectively compromised in their in-room fidelity and subjectively "dull."

With that said, I appreciate the comments already provided here by @amirm, @NTK, and @olieb, among others, which help explain and clarify some of the issues, including the negative DI in the lower bass region (which Arindal has elewhere claimed is evidence for inherently "bloated" bass that does not otherwise show up in any measurements).

[* I wrote nonspecifically of "increasing DI in a given range" because in various posts Arindal has cited ranges of 500Hz-2.5kHz, 800-5k, 800-6k, 800-8k, and 400-800Hz. So it appears to be all over the place with the range seemingly changed to correspond with the DI of whatever speaker he's applying this claim to at a given moment.]

There's too much attention given to DI. It's a single composite curve. Why bother when you can compare maps or groups of curves, which far better illustrate radiation behaviour even in cases where some amount of error is to be expected because of the measurement technique (thinking of windowing).

@Arindal's claims seem like nonsense to me. If I take them seriously they suggest there are major flaws with existing research. I'd like to see evidence.

 

[Arindal]

There's too much attention given to DI. It's a single composite curve.

I think there is not enough attention given to DI. Yes, it is an averaged curve, and it does not allow precise predictions of sound quality and character. But it explains what has most certainly gone wrong when people are dissatisfied with tonality or imaging in the room.

My personal opinion is there is way too much attention given to on-axis response. It is the only thing you can easily reverse in a speaker with the simple use of DSP.

@Arindal's claims seem like nonsense to me. If I take them seriously they suggest there are major flaws with existing research.

Is the unwillingness to doubt ´existing research´ the only reason to dismiss the theory? Or have you made an A/B comparison yourself? I mean, which research of the last 15 years (since competitive constant directivity speakers are available as potential A/B test rivals) has underlined the theory that colorated reverb is preferable over linear reverb tonality?

I had the chance to have a blind A/B comparison by being provided with otherwise identical speaker prototypes, allowing a comparison of solely different tweeter directivity while the on-axis response stayed the same. The experiment was repeated several times by recording engineers and students. It was pretty interesting and made me seriously doubt the positive outcome of waveguides, which I had previously preferred.

On the other hand, I am not doubting the findings by Toole regarding preference. I don't think there is a major flaw in the research, but that the way the experiments had been conducted at Harman lab back in the days, were leading to a result which does not translate properly to the world of living room listening, people's preference with their own favorite music and the way they set up their speakers. Interestingly, AJ in the stereophone interview refers to Toole´s preference experiments:

And so when we look at all of Floyd Tool's work .. what's got to be understood is the work of relating measurements to listening was done on a preference basis. They play recordings that play them over different speakers, make all the measurements of the speakers and then try and work out um what people prefer to hear. Well, the majority of people prefer Yes.
And all of them. Yeah. Some people prefer this band, some people prefer that. And so you could categorize them into that. But it was still based on commercial recordings. It's not like they were had a band in the next room or let's say a string quartet and they captured the sound of it and then play it back over these different speakers and you can go between that room and this room and say,
"Oh, this one sounds most like that." They didn't do that. It's commercial recordings and it's based on preferences.

He might have nailed what looks like a contradiction. Maybe there is none, it is just the personal subjective preference of people being asked ´do you like speaker A or speaker B or speaker C the most?´ with commercial recordings (in particular popular recordings) unkown to them, in a lab where they don't see the speakers and under acoustic conditions they are unfamiliar with.

 

[tmtomh]

To claim that Andrew Jones, one of the world's best speaker designers, doesn't understand measurements, their importance, mathematics, physics, and their application in acoustics, I find it risky. Perhaps we should measure less and listen more.

I agree with you that it is "risky," as you say, to claim that Andrew Jones doesn't understand measurements or their implications.

At the same time, I think the problem with the linked article that is the subject of this thread is that the author (or the AI that wrote it) is taking Jones' quotations and using them in such a way that it implies Jones agrees with the specific issues cited in the article. I don't think that's the case - in other words I don't think Andrew Jones himself would agree with everything in that article, even though it's alleged to be said under the authority of his name. For example, the item I raised earlier in the thread: one of the three major alleged problems with measurements is said to be the mixing-together of speaker performance and room effects. But the entire article is framed as a critique of the "Klippel analyzer," when in fact the Klippel system that's used to measure speakers is specifically designed to avoid precisely that issue.

So it's not about saying Andrew Jones doesn't know what he's talking about. It's about saying that article doesn't know what it's talking about.

And finally, I don't think your final sentence follows from your two prior ones. Of course listening is important - and of course measurements are important. If there is a general tendency in the audio world, it is towards too much reliance on listening and not enough on measurements - although again I agree that of course we must listen, as that's the ultimately point of it all.

 

[Arindal]

Typically diffusers in small rooms are positioned on the front wall, back wall or ceiling, not the side walls. The aim is for later arriving reflections to be scattered in time.

Have never met an acoustician following such concept in a room dedicated to reproduction. Usually diffusors on the side walls are the first measures in a small room, as discrete early reflections from the sides are most likely to cause deterioration of localization and give our brain the image of a small room. Front wall treatment is in many cases obsolete, and ceiling and back wall might be necessary if their typical flaws were identified in listening tests.

If there is a general tendency in the audio world, it is towards too much reliance on listening and not enough on measurements

I think we agree on this point. The interesting question is: Why are people in the audio world relying decreasingly on measurements, as they are getting better and better, becoming more publicly available (Kudos to Amir, Erin and JA!) and being prominently promoted by sites like ASR? It is kind of an irony.

 

[tmtomh]

I think we agree on this point. The interesting question is: Why are people in the audio world relying decreasingly on measurements, as they are getting better and better, becoming more publicly available (Kudos to Amir, Erin and JA!) and being prominently promoted by sites like ASR? It is kind of an irony.

I'm glad we agree that the general audio-hobby tendency isn't weighted enough towards measurements. But I can't agree with the final question you pose, because I think it's based on an empirical claim ("people in the audio world [are] relying decreasingly on measurements") for which no evidence has been provided.

I know ASR is a bubble, but these days so is almost any online community, and ASR is a pretty big bubble, to gauge by the reported site traffic (especially in comparison with subscriber numbers of the legacy audiophile publications), and it's obvious ASR and other measurement-based venues are having an impact in recent years. So while I am certainly open to the idea that ASR's impact is not as large as we might think it is, I don't see that the reliance on measurements is actually decreasing in the audio world.

Or are you trying to imply that because several members have disagreed with your very specific claims about the relationship between the DI measurement and your and other listeners' impressions, that means many ASR members are "relying decreasingly on measurements"? Hopefully that's not your argument, because that argument would be both unpersuasive and tendentious.

 

[Heinrich]

... Arindal's claim that speakers whose DI increases by more than 2-3dB over a given range* ...are both objectively compromised in their in-room fidelity and subjectively "dull."

... Arindal has elewhere claimed is evidence for inherently "bloated" bass that does not otherwise show up in any measurements).

[* I wrote nonspecifically of "increasing DI in a given range" because ...

Thanks.
a) negative DI in bass, as I explained before is an artifact from evaluating / processing the raw data of NFS for a close virtual single listening point. It must not be present in true far field.

b) once and again the sanctuary DI. The DI is one, room acoustics is the other. The two work together hand in hand. The floor is covered with a carpet, a thick Nepalese or is it a skinny silk-made. No carpet, aha ... . A shelf with real books at the side or front/back, elsewhere etc? And where are the most relevant ears located, exactly?
People always tell me, that "PIR" would replicate itself at their individual homes. For one, at +/-3dB or so at best, second, not really. The "PIR" refers to a certain measurement condition that many might not know about, let alone sticking to it. It's in the standard.

It is not the data, but how it is read. Very much like with other audio data before the use gets out of hand, ending up in some inappropriate obsession. So far I'm in alignement with the article in question. The first part is, in my book, quite a bit off.

 

[tmtomh]

Thanks.
a) negative DI in bass, as I explained before is an artifact from evaluating / processing the raw data of NFS for a close virtual single listening point. It must not be present in true far field.

b) once and again the sanctuary DI. The DI is one, room acoustics is the other. The two work together hand in hand. The floor is covered with a carpet, a thick Nepalese or is it a skinny silk-made. No carpet, aha ... . A shelf with real books at the side or front/back, elsewhere etc? And where are the most relevant ears located, exactly?
People always tell me, that "PIR" would replicate itself at their individual homes. For one, at +/-3dB or so at best, second, not really. The "PIR" refers to a certain measurement condition that many might not know about, let alone sticking to it. It's in the standard.

It is not the data, but how it is read. Very much like with other audio data before the use gets out of hand, ending up in some inappropriate obsession. So far I'm in alignement with the article in question. The first part is, in my book, quite a bit off.

Thanks for your reply (and for what it's worth, your prior comments were at front of mind when I added "and others" to the three names of folks whose contributions to this thread I have found very heplful - so thank you!)

In fairness to Arindal, I believe he has argued that in the near field or in treated rooms the DI measurements he doesn't like are not necessarily a problem. That said, I find that kind of backwards, because by that logic he shouldn't be saying narrower-directivity speakers are okay in nearfield or treated spaces, but rather he should be arguing the converse: he should be claiming that we should all be listening in farfield in minimally treated rooms, so as to allow the full bloom and impact of the off-axis reflections of speakers with DI he likes.

And this is my problem with his argument: at the core of his argument, the wider vs narrower directivity preference seems to magically cross the Rubicon from (a) listener preference to (b) objective fidelity measurement, simply because he says it does.

 

[Heinrich]

I don’t get @Arindal’s argument. One thing: to tell you're listening in the near field says nothing about the room. In a big room, uneven radiation in different directions will get mixed and damped, while early reflections in small rooms pick out certain radiation angles. Because of stuff like this, a deep dive into NFS data can totally miss the point if you actually care about realism. Sure, a smooth off-axis response across all angles makes a speaker more “universal,” but whether and how far that actually helps in a real setup is another question.

I gotta stop here because these made-up complexities are wrecking my already terrible English. If there were actual data, or a data-driven model, none of this would be a problem. But I don’t see that yet as a useful complement to the speculations being thrown around.
I wish the often-cited literature wouldn’t just be cherry-picked to support these ideas, but used as motivation to actually try something yourself. Small, systematic investigations, even if it’s not big science right away.

 

[Toni Mas]

I agree that asking a panel of listeners to vote what speaker they liked most the way they render a few top of the pops and make stats with this stuff have more to see with typical marketing studies part of new product development than a strictly scientific approach of loudspeaker design.

 

[Keith_W]

Based on the "Why do some people say Genelecs are too flattering or forgiving" thread, I was about to start a new thread dedicated specifically to getting knowledgeable members' input on the questions of negative DI in the lower bass region and DI-increase slopes throughout the lower mids to lower treble, because I'm always open to the possibility that maybe I'm just crazy.

Not sure what you mean by "negative DI in the lower bass region". You know that DI's are normalized so they don't disappear off the bottom of the graph, right? The calculation is 10 * log10 (on axis/sound power).

 

[amirm]

To claim that Andrew Jones, one of the world's best speaker designers, doesn't understand measurements, their importance, mathematics, physics, and their application in acoustics, I find it risky. Perhaps we should measure less and listen more.

There is zero requirement for even the best designers to understand the mathematics of Klippel NFS. The people in the industry who do understand such things, such as Bruno Putzeys, have papers and presentations that are full of math. We have none of that for Andrew Jones. Andrew is also a practical designer, not one that knows the physics to perform simulations such as our resident @René - Acculution.com. I routinely read statements that are flat out wrong about operation of Klippel NFS, and even measurements I perform on AP from designers.

I see this all the time where people assume that someone who designs audio products by definition is an expert end to end in all things audio. This is not true at all. Take acoustic design. There are people who say they know how to design a room. But have you seen this kind of simulation from them, comparing and optimizing local of one sub vs multiple?

What you are seeing is the simulation that was performed in Comsol by a PhD researcher working for Keith Yates. There is $100,000 worth of hardware and software plus a ton of knowhow to create simulations like that.

I am fortunate enough to have had a leg in many domains so can converse with subject matter experts in all of these areas. It is that knowledge that allows me to easily judge whether a designer has the expertise we are talking about, or not.

 

[amirm]

People always tell me, that "PIR" would replicate itself at their individual homes.

No one should be saying that. PIR is a standardized method of looking speakers in a hypothetical room that is constant between speakers. It is a rationalized method of mixing on and off axis response. Without it, it is hard to integrate the off-axis into on-axis visually. Due to averaging it performs, it also tells you what is a more serious variation which survives that averaging.

In your own home, your job is optimizing low frequencies. And tailoring a target response to taste. If you have to screw around in higher frequencies, then you have bought the wrong speaker. Which our measurements show.

 

[amirm]

The interesting question is: Why are people in the audio world relying decreasingly on measurements, as they are getting better and better, becoming more publicly available (Kudos to Amir, Erin and JA!) and being prominently promoted by sites like ASR?

We have vastly moved the needle with respect to importance of measurements in the entire spectrum of audio products. The effect varies in category but it is 100% there or we wouldn't be have the retorts that started this thread. The impact is so big that folks are feeling threatened, trying to throw darts at measurements and creating talking points.

 

[a4eaudio]

Not sure what you mean by "negative DI in the lower bass region". You know that DI's are normalized so they don't disappear off the bottom of the graph, right? The calculation is 10 * log10 (on axis/sound power).

I assume he means by "negative DI" when the directivity index goes below zero (i.e., negative).
And if the graph was limited to a lower axis of zero it would in fact "disappear off the bottom of the graph". As seen below...

 

[olieb]

Not sure what you mean by "negative DI in the lower bass region". You know that DI's are normalized so they don't disappear off the bottom of the graph, right? The calculation is 10 * log10 (on axis/sound power).

log10 (on axis/sound power) becomes negative when on axis is smaller than sound power [actually it is the averaged spl over the (full) circle or sphere] and therefore the ratio in the logarithm is smaller than 1. This happens in many cases in very low frequency.
One example is the mentioned Genelec 8361.
From spinorama.org (as in the ASR review the graph is cut off at the bottom).

The reason in this case being the port in the back. Below 50Hz it is the dominating source of sound and the waves have to travel an extra 50-60cm (compared to the sound from the drivers) to the mic at 2m in front of the tweeter. This will result in about 2dB less sound pressure at the mic compared to the average radiated into the room. Therefore DI ≈-2dB

Just noticed that @a4eaudio beat me to it.