The ASR "objectivists tribe" ;) has to read the Keynotes at the AES160th by Lars Risbo

[RogerSmith]

He makes many bold and, as seems, unsubstantiated claims. For example he say "The exponential sine sweep, the standard method we use to generate THD plots, works by spreading harmonics out across time. The harmonics of the sweep arrive at the listener’s ear at different times. Temporally separated like that, people routinely detect distortion products even at minus 80 decibels." Is there any evidence or experiment where somebody could detect distortion at minus 80dB? I seriously doubt it.

 

[amirm]

what exactly are those hyper relevant measurements we omit?

They are driver level measurements they performed to minimize distortion.

 

[MattHooper]

How much authenticy do I want? So often, when attending live concerts (which I do regularly, I have a subscription to Symphony Hall and attend several other concert venues...) the question is... do I want that exact sound at home. My answer would be "no".

Essentially, I feel the same way.

Which might be odd since I have so often talked about my fascination with live versus reproduced sound, and how I’m trying to re-create certain aspects of real sound sources with my sound system.

But ultimately, I’m not looking for perfectly realistic reproduction. In my home theater for example, I certainly do not want the perfectly authentic sound of guns, tanks, grenades and mortar fire going off beside me when I’m watching a war movie. Obviously it would be insanely uncomfortable, and my hearing would be shot in no time. I just want certain aspects of the movie watching experience to allow me to sink into an illusion. (for instance if I was knowledgable about war, I would want to see the right artillery being used for the right time period).

Likewise, I don’t want the sound levels and pressure of Neil Peart’s gigantic drum set being played full volume in my small listening room. I’m never going to be trying to re-create that volume. But what I do want are certain elements I associate with the real thing - for instance and authentic recognizable sense of timbre to the snare drum skins, cymbals….

I’m just looking for elements that make for a comfortable, engaging illusion.

And speaking of that….

Regardless of what we do with reproduction, we're not going to improve on the source material, and if the lead vocals were recorded on an SM-58, you can't tell me that a $50k speaker is going to make it any better.

I think it depends on what one sees as an improvement, and the goal one has for a sound system.

To my ears something like the MBL 101E Omnis (now well over $50,000) represented an advance in sonic realism. I’ll never forget first encountering them at an audio show where they blew my mind… to me they made everything else at the show, and every other conventional speaker I’d heard, sound like music being squeezed out cones and tweeters.

Those are never going to be inexpensive to build.

 

[pablolie]

They are driver level measurements they performed to minimize distortion.

Are they hyper relevant, though? Would additional "relevant" measurements demands ever stop to convince the crowd that believes you can't measure "real" audio output? Does cabinet resonance need to be measured to support nuances of string instruments? It would never end, I think.

 

[Buckchester]

Are they hyper relevant, though? Would additional "relevant" measurements demands ever stop to convince the crowd that believes you can't measure "real" audio output? Does cabinet resonance need to be measured to support nuances of string instruments? It would never end, I think.

Probably relevant to a degree. But to what degree...probably less than what we already have in a spin IMO.

 

[JiiPee]

In the end, doesn't it come back to the simple question: Can one detect it in a controlled blind test?

If the answer is no, then there is still room for further research and development for those who are interested in technological excellence per se, whereas those who see the technology as a tool which is either good enough, or not enough, for the task it was created for, can concentrate on other aspects of audio equipment like look and feel etc...

 

[MattHooper]

Interesting paper.

I do find that the two tribes he described were pretty strawman like - at least his description of the objective tribe’s claims.
But I guess, he was looking for a sort of exaggerated illustration on which to base his argument.

I’m not technically literate enough to weigh in on many of the claims, but some thoughts did arise especially in response to these passages:

“One mechanism we focused on was magnetic hysteresis distortion in loudspeaker motors. You may have heard this in a system without knowing what it was. A granular texture in the sound.

A blanket of fuzz that stays just this side of audible — taunting and infuriating, like an itch you cannot scratch. That is hysteresis.”

I recognize that perception!

I have long perceived that teeny bit of graininess/fuzziness when listening to the vast majority of sound systems. In many instances I don’t know the cause of that perception - whether it’s fully my imagination, or something objective in the sound causing that perception, or some mix of both. (I do know that sometimes it is simply caused by higher frequency room reflections - “room hash” as I tend to call it - which can be mitigated by cutting down those reflections).

This type of perception is related to what audiophiles often refer to as a “ black background.” The sense that some overlaying spray of distortion, fuzz or grain has been removed allowing the space between the notes and instrument to become the equivalent of “ deep black” which has a sort of calm effect as well as allowing for instrumental timber to sound more pure and vivid. Similar to a high-quality CRT display in which some interference noise and speckles have been removed allowing for the deepest black and the picture to have maximum contrast and vividness and richness.

I think that the whole “black background” description by audiophiles tends to be unreliable and undemonstrated. But whether somebody appealing to the concept happens to be describing a real audible phenomenon in a system or if their perception of a black background is their imagination, I understand what they mean perceptually… as I have sometimes experienced it myself (it’s actually one of the reasons I bought my current loudspeakers because of that perception).

So I found it interesting to see the author mentioned something that I seem to perceive very often. Whether it’s actually ever to do with hysteresis I don’t know, and I would remain sceptical unless there’s more rigorous data in support of the idea.

On Toole’s work:

“Were the test speakers representative of what is now achievable? Is listener preference even unimodal — or are we averaging over genuinely different clusters whose mean satisfies nobody?”

Unless I’m missing something… isn’t that question already answered by the fact that Toole et al tell us that the preference/measurement correlation models were able to have very strong predictive power? That they predicted with a very high likelihood preferences for certain loudspeakers? And wouldn’t this apply to individuals as well as groups? (I don’t mean perfect predictability, but strong enough predictability to answer the author’s question?)

“The problem is not what Toole claimed. The problem is what accumulated around his work over time — independently of him.”

I’ve certainly seen some of this. Occasionally some of Toole’s defenders are less appropriately cautious in their inferences from the studies and data than Toole.
And sometimes they seem to imagine that simply citing the research or citing Toole amount to some open and shut case for their own arguments, when they don’t seem to be aware of the role of their dubious interpretation of the data or arguments. “I’m citing the science so if you disagree with my argument you disagree with the science. Therefore, in disagreeing with what I write you must be anti-science! ”

There’s data and then there is what can be inferred from the data.

 

[Mr. Widget]

I think everyone here in the town hall of "the objectivists tribe" has to read the great keynote by Lars Risbos.

Unfortunately reading it alone isn't enough. One needs to think about what he is saying. He brings up many interesting and good points. We can all take exception to this point or that, but on the whole, it is a very thoughtful study of our industry (or hobby).

The most obvious quibble is the two tribes notion. I do agree with his point that in audio as in many other areas of our lives we are becoming tribal, but there is a lot of nuance missed by lumping everyone into the objectivist or subjectivist camps.

 

[Leeken]

Just how many well performed peer reviewed studies have there been into audibility with regard to music reproduction equipment? (In particular speakers) Pretty sure Tooles research is? Is there many more other than that? Electronics is obviously very easy to set up,most of us can do that with pretty high confidence in our own home,but speakers aren’t easy even just to do a proper A-B,never mind blind abx etc etc

 

[Duke]

@test1223, thank you very much for making me aware of that article.

This is one of the passages that caught my attention:

"when a study reports a result is not statistically significant, that does not mean the effect is absent. It means the study failed to reject the null hypothesis — which could mean the effect isn’t real, or it could simply mean the study was not powerful enough to detect it.

"“Not proven audible in this study” is routinely translated into “not audible.” That translation is wrong. Absence of evidence is not evidence of absence. We say this but do not act like it."

On another forum I recently had a conversation with researcher Earl Geddes. One of the topics had to do with the effect on imaging of a strong early same-side-wall reflection, versus the effect on imaging of the somewhat-later-arriving opposite-side-wall reflection. Earl and Lidia Lee had conducted a study of the subject which they ended up not publishing. Here is his post, copied-and-pasted:

"Dr. Rumsey's comments remind me of an unpublished study that Lidia and I did looking at near wall versus far wall reflections. We asked listeners to define the phantom source locations in the two different configurations. We did not publish because the results were not stable enough for Lidia to agree to publish. (The task was too difficult for the untrained listeners.)

"However, to me, it was clear that a far wall reflections offered a much clearer localization than a near wall reflection. This is critical to understand another rational for why I toe-in my speakers: to avoid a near wall reflection even though it likely increases the far wall reflection - better imaging."

This is like the "not proven audible" case that Lars describes, in that their finding of "not proven audible" was not equivalent to "inaudible".

So you can have people (like me) advocating for something unorthodox (speakers + setup geometry which illuminates the far wall instead of the near wall), and you could easily say "show me the proof" and scoff when "the proof" doesn't exist (because in this case the study that looked into it failed to find proof to a statistically reliable degree). But the unorthodox idea's status of "unproven" is not the same thing as the unorthdox idea being "disproved".

At least, I think that's the point Lars was making.

I don't think there is anything wrong with taking an approach that values the proven over the unproven, but if there is progress yet to be made, imo it may involve intelligently delving into that which is currently unproven. And imo that is the very reason for someone in the industry to familiarize themself with that which has been proven: So that if they choose to step into the realm of the unproven, they do so intelligently.

As for the two tribes thing, here's my take: Yes there are two tribes. They are: Those who divide everyone up into two tribes... and those who don't.

 

[voodooless]

I’m sorry guys, I think most of you are way too charitable here. I don’t see how this article (why do people call this a paper?) is so much different from much of the other audiophile woo we regularly dissect and criticize. It’s the exact same structure that for instance Chord uses to try to sell you their next M-scalar:

1. start with real engineering
2. point out that ordinary measurements or research have limits
3. suggest there is an important missing variable
4. imply their product/design addresses it
5. stop before providing direct evidence that the alleged missing variable matters audibly in normal use

.. Add some strawmaning of your opponents positions. And obviously, #4 is implied here, you can’t go about selling your shit at an AES keynote… it’s like claiming the Eurovision isn’t political

As to the actual arguments: the IDT stuff is just silly. It’s only seen in very specific controlled experiments, and they all involve headphones, not speakers in a room. There is zero evidence that this translates to real world scenarios.

The THD and Toole arguments are just straw men: nobody here that really understands these topics uses such a simplistic interpretation. And then there is some
more conjecture stacking.

Don’t get me wrong, Purifi makes outstandingly excellent products, the pinnacle of current transduction engineering! And I’m also sure we still have quite a bit to learn about loudspeaker development and psychoacoustics. But then come with some actual examples that are grounded in reality, with some actual audibly proven effects, not a bunch of conjecture. If you spend so much time on your engineering, make sure that you can back up why you did it. This is what will make you stand out against all the other audiophile companies.

 

[kemmler3D]

there is a lot of nuance missed by lumping everyone into the objectivist or subjectivist camps.

Yes, but the people who don't reflexively align with a camp and trade in nuance implicitly escape being lumped into a tribe, I guess.

 

[Keith_W]

I believe that Lars Risbo is correct. He is essentially arguing that the field of psychoacoustics is underdeveloped, and many of our questions have not been answered. I too am FAR from a psychoacoustician, but I have read the papers. What I saw were a whole bunch of studies that try their best, but ultimately do not answer relevant questions. One example: we can not extrapolate studies done with headphones to listening rooms, yet we routinely do that. Not because we are stupid, but because we have no choice. There may be NO relevant studies done in listening rooms, as Dr. Risbo points out.

He is also correct to point out that the masking studies were done with simultaneous tones. Are the tones still masked if they are temporally separated? And if they become unmasked, what is the temporal threshold? Because, music does not just consist of tones, they also have transients. What then, if transient smearing separates the tones? Once again, nobody knows the answer because the studies have not been done!

My take-aways from that keynote:

1. Stop being so bloody dogmatic! Be more open-minded about what may be audible, and what may not be audible.
2. Be careful about what the studies say, and more importantly: what they don't say. Do not place too much weight on secondary endpoints. The study was not designed to look at those, and drawing conclusions from it may be misleading.

These are all points I wholeheartedly agree with. Dr. Risbo might call himself an engineer, but he is no mere engineer. He has proven that he understands science more than any of those people on ASR who are complaining about his talk. Let alone some of you who have clearly not understood what he said or skimmed the article before bloviating your half-baked thoughts on ASR. You are an embarrassment - when I see that, I sometimes feel like youngho and wonder why I am part of this community (the answer is, there is nowhere else to go ... otherwise I would have gone there ages ago!).

 

[Mort]

and wonder why I am part of this community

This is true of all of us, I think. y'all are a bunch of annoying weirdos.

 

[Karmacoma]

5. stop before providing direct evidence that the alleged missing variable matters audibly in normal use

Amen to that. And more specifically he tries to evade the burden of the proof by basically saying people (who don't agree with them) are narrow minded.
That's not how it works, you need to provide enough data and a reproducible experiment so everybody can try and criticize your theory at lenghts until it's considered solid enough to reach the consensus level, you can't skip that and complain about being criticized, even if it's long and costly, that's the scientific process.
But yeah he came for a keynote, he made a keynote, can't blame him too much on playing his game.

 

[voodooless]

My take-aways from that keynote:

1. Stop being so bloody dogmatic! Be more open-minded about what may be audible, and what may not be audible.

But as always: don't let your brain fall out!

2. Be careful about what the studies say, and more importantly: what they don't say. Do not place too much weight on secondary endpoints. The study was not designed to look at those, and drawing conclusions from it may be misleading.

And yet, this is exactly what he does. And this is exactly the criticism I have!

He says:

• Don’t overread studies
• Don’t lean too hard on secondary endpoints
• Don’t draw conclusions a study was not designed to support

And then he does exactly that with:

• ITD work
• distortion audibility in sweep contexts
• crossover/group-delay hints
• room-mode reinterpretations
• and Toole-adjacent preference research

Let alone some of you who have clearly not understood what he said or skimmed the article before bloviating your half-baked thoughts on ASR. You are an embarrassment

Yeah, yeah, let's stop the name-calling and get the actual arguments on the table, please!

 

[Frank Dernie]

I think of it like race cars. Lap time predicts race performance pretty 1:1 but it's still worth studying tire temperature, suspension behavior, or aerodynamics. Toole's research is lap time but there may still be unexplored contributors to performance.

Being pedantic it is much more complicated than that, but I agree that there are likely to be contributary factors many, if not most, people are unaware of.
Toole's research shows pretty clear preferences and it has stood the test of time.
OTOH it is preference and, for example, on a much smaller level, would 2 speakers with the same radiation pattern and even frequency response be audibly distinguishable by distortion, for example.

FWIW lap time on a single lap is a not a patricularly good determinant of race performance, it takes many years to learn this, and what is actually key in the race and, IME, even a significant number of people involved don't know all the crucial things, mainly because nobody who knows has written a technical book about it, and young engineers are hard to convince about things they haven't seen in a text book and kept secret by those that do know. It probably took me 20 years to really understand the most important aspects, and I have never seen them written down
Plus there are probably things I didn't do it long enough to learn...

 

[voodooless]

would 2 speakers with the same radiation pattern and even frequency response be audibly distinguishable by distortion, for example.

I find this example always so enlightening:

BLINDTEST: Midrange 360-7200hz, NO audible difference whatsover.

Just realized no proper thread was made to explain in a simple way the blindtest I organized a little while ago: So here it is: Blindtest. Rotating cube (enclosure) with 2 different drivers per round, one driver at a time, on-axis to the listener. Bandpass crossover 360hz @ 7,200hz (48db/oct...

www.diyaudio.com

 

[Mort]

My two tribes are the people that can't argue without making a car analogy, and those who can.

 

[solderdude]

For me the real take-aways were:

The subjectivists tribe: audio is art, measurements are irrelevant, no explanation is too absurd provided it avoids actual data.

The objectivists tribe: standard metrics are sufficient, double-blind testing is the only admissible evidence, and almost nothing is audible. And yet — paradoxically — they spend enormous effort optimizing the same short list of metrics: THD, IMD, noise floor. Decade after decade. Regardless of whether those metrics capture what the ear is actually sensitive to.

What bothers me about both tribes is not that they disagree. It’s that neither is curious. Both have found a way to stop the inquiry and call it a conclusion.

I do believe in both tribes there are folks that are curious but in the end are not willing to switch tribes.

“Not proven audible in this study” is routinely translated into “not audible.” That translation is wrong. Absence of evidence is not evidence of absence. We say this but do not act like it.

This is demonstrated by demanding 'proper blind test results' in situations where it is really hard to do so ...properly.
Audibility thresholds (of each individual) is what it is all about.

Audio is not art and it is not science. It is engineering.

O.K. it is time for another Dutch audio-wisdom-tile.