A Broad Discussion of Speakers with Major Audio Luminaries

[Jaxx1138]

You apparently don't know what the circle of confusion is.

Floyd Toole's "Circle of Confusion" is a concept that highlights a fundamental problem in the audio industry regarding the consistency and accuracy of sound reproduction.

Essentially, the "Circle of Confusion" describes a self-referential loop where:

1. Recordings are made: Sound engineers create recordings using microphones, processing (EQ, compression, effects), and mix them.
2. Monitoring is used: These engineers evaluate and make decisions about the recording by listening through a set of monitor loudspeakers in a specific room.
3. Loudspeakers are designed: Loudspeaker manufacturers design and evaluate their loudspeakers by listening to recordings.
4. The loop continues: These recordings, in turn, were made using microphones and mixed through monitor loudspeakers, which themselves were designed by listening to recordings, and so on.

The core issue is the lack of a standardized, calibrated monitoring environment including all equipment (loudspeakers and electronic that meet the defined standards)

Several of the issues:

• Inconsistent Recordings: Recordings can vary dramatically in their spectral balance, dynamic range, and spatial imagery because engineers are mixing them based on what sounds "right" on their potentially flawed or uncalibrated monitoring systems. For example, if a studio's monitors are deficient in bass, the engineer might unknowingly boost the bass in the mix to compensate, leading to a bass-heavy recording when played on a neutral system.
• Difficulty in Evaluating Components: For consumers and reviewers, it becomes nearly impossible to judge the true accuracy of an audio component (like a loudspeaker) when the recordings used for evaluation are themselves an unknown variable. A "bright" recording might make an accurate loudspeaker sound too bright, or make a "dull" loudspeaker sound good.
• The "Lie" of Recordings: As some engineers have put it, "All recordings are lies. The best recording/mix engineers are the best liars." This doesn't mean deceit, but rather that a recording is a creative artifice intended to convince the listener they are present at a live event, not a literal capture of it. The lack of standards exacerbates this artificiality.

Floyd Toole argues that breaking this "Circle of Confusion" is crucial for advancing the quality of audio. He emphasizes the need for:

• Standardized, calibrated monitoring environments in professional studios.
• Loudspeakers that measure and sound neutral across a wide range of listeners and rooms, based on scientific research and double-blind listening tests.

By creating greater consistency and accuracy at the production end and promoting more accurate reproduction systems for consumers, the goal is to ensure that listeners hear the audio "art" as consistently and accurately as possible, reducing the "confusion" introduced by varying interpretations and flawed equipment throughout the audio chain.

 

[MattHooper]

Floyd Toole's "Circle of Confusion" is a concept that highlights a fundamental problem in the audio industry regarding the consistency and accuracy of sound reproduction.

Essentially, the "Circle of Confusion" describes a self-referential loop where:

1. Recordings are made: Sound engineers create recordings using microphones, processing (EQ, compression, effects), and mix them.
2. Monitoring is used: These engineers evaluate and make decisions about the recording by listening through a set of monitor loudspeakers in a specific room.
3. Loudspeakers are designed: Loudspeaker manufacturers design and evaluate their loudspeakers by listening to recordings.
4. The loop continues: These recordings, in turn, were made using microphones and mixed through monitor loudspeakers, which themselves were designed by listening to recordings, and so on.

The core issue is the lack of a standardized, calibrated monitoring environment including all equipment (loudspeakers and electronic that meet the defined standards)

Several of the issues:

• Inconsistent Recordings: Recordings can vary dramatically in their spectral balance, dynamic range, and spatial imagery because engineers are mixing them based on what sounds "right" on their potentially flawed or uncalibrated monitoring systems. For example, if a studio's monitors are deficient in bass, the engineer might unknowingly boost the bass in the mix to compensate, leading to a bass-heavy recording when played on a neutral system.
• Difficulty in Evaluating Components: For consumers and reviewers, it becomes nearly impossible to judge the true accuracy of an audio component (like a loudspeaker) when the recordings used for evaluation are themselves an unknown variable. A "bright" recording might make an accurate loudspeaker sound too bright, or make a "dull" loudspeaker sound good.
• The "Lie" of Recordings: As some engineers have put it, "All recordings are lies. The best recording/mix engineers are the best liars." This doesn't mean deceit, but rather that a recording is a creative artifice intended to convince the listener they are present at a live event, not a literal capture of it. The lack of standards exacerbates this artificiality.

Floyd Toole argues that breaking this "Circle of Confusion" is crucial for advancing the quality of audio. He emphasizes the need for:

• Standardized, calibrated monitoring environments in professional studios.
• Loudspeakers that measure and sound neutral across a wide range of listeners and rooms, based on scientific research and double-blind listening tests.

By creating greater consistency and accuracy at the production end and promoting more accurate reproduction systems for consumers, the goal is to ensure that listeners hear the audio "art" as consistently and accurately as possible, reducing the "confusion" introduced by varying interpretations and flawed equipment throughout the audio chain.

One big problem: even if tomorrow we woke up in the world where suddenly consistent standards had been adopted in recording studios and consumer playback, we still have the problem that the vast majority of recorded music has not been produced under such situations. And so if you care about music already produced, you’re always going to be stuck with the circle of confusion.

For now that leaves us with trying to set up systems that allows us to enjoy our music as much as possible. Which is going to involve personal preference.

And of course the type of blind testing you were involved with had to do with establishing preferences, not accuracy per se.

And interestingly, irrespective of whether the circle of confusion is ever broken, it still seems we have good reason to have
“ accurate” playback on one side of the circle - the consumer side. Because it turns out that the sound of accurate systems - neutral, low colouration - tends to be preferred in of itself by most people (as established by the research cited/conducted by Toole et all).

And it would seem that a preponderance of recorded music - even including all the variables in how it is made - can sound good on such systems. (Otherwise we would continually having to reach for tone controls or EQ for every different track or album we play, which would be a chore).

(Though I certainly believe all the above is legitimate reasoning for somebody seeking as accurate playback as possible, I also don’t think this rules out flavouring a system to ones taste overall… since purely accurate playback doesn’t get us out of the circle of confusion, so as long as one is enjoying a preponderance of music through one’s music system… whatever floats your boat).

 

[Arindal]

we still have the problem that the vast majority of recorded music has not been produced under such situations.

Certainly true. But even if we accept that millions of recordings have been mixed and mastered on different systems, I would say there are a lot of similarities between studio control rooms, and the variations between most of them is much smaller than found between typical hi-fi setups at home. So if precise standardization cannot be achieved, we can still do some things to stay within the usual tolerance band of deviations found in studios. Listening in stereo and keeping and equilateral stereo triangle is the single most standardized aspect, while supressing disturbing early reflections achieving stable center phantom localization being the second most common. So why not follow at least these two?

For now that leaves us with trying to set up systems that allows us to enjoy our music as much as possible.

Yes. But how do you define which of your favorite tracks gives you the most of enjoyment from sonic point of view? If your system deviates from the average studio tolerance band, one recording might sound fabulous, the next one annoyingly kinked. To leave the circle of confusion, I plea for taking the percentage of recordings which ´just sound right´ and give enjoyment if they are within the genre limits of the listener´s taste, as measure how well the average sound reproduction is met.

If an audiophile tells you ´only the 10 best audiophile recordings sound superb on my system´, you have an estimate how far it must be off the standards. On the other hand, something between 80-90% of degree of satisfaction is achievable.

The other method of leaving the circle of confusion is visiting live concerts without sound reinforcement. From my personal point, the single most reliable way of judging if a system in a room ´sounds right´ is to visit a concert which will be subject to a broadcast or album issue later on. Knowing both, preferably with little delay between both listening sessions, gives you an astonishingly good understanding of how things should sound and what the mixing engineer was meaning. I personally prefer to hear what the broadcast mixing engineer is doing during general rehearsal, before I visit a concert, but I understand that is not doable for everyone.

 

[youngho]

The other method of leaving the circle of confusion is visiting live concerts without sound reinforcement. From my personal point, the single most reliable way of judging if a system in a room ´sounds right´ is to visit a concert which will be subject to a broadcast or album issue later on. Knowing both, preferably with little delay between both listening sessions, gives you an astonishingly good understanding of how things should sound and what the mixing engineer was meaning. I personally prefer to hear what the broadcast mixing engineer is doing during general rehearsal, before I visit a concert, but I understand that is not doable for everyone.

I tend to consider this as a cognitive illusion, since concertgoers can have different seating location and sonic preferences (see comments by Beranek and Lokki at https://www.audiosciencereview.com/...cert-hall-acoustics-links-and-excerpts.51487/), and virtually none of these seating locations correspond with typical recording microphone locations (https://www.psaudio.com/blogs/coppe...6XAzCGb2Lq2lf2zmJU7KXjcgK9NvfRHKNrYpUxWyU5Wuz), and the vast majority of us can only presuppose the engineers' intents without direct conversation or published description, even with which we must assume that description of intent completely and accurately captures the actual intent, while cognition typically involves at least two systems, one of which can occur below the level of awareness. With respect to the latter link, some criticized Kavi Alexander's Water Lily recordings of classical music as being too dull or muffled while others felt like it did come closer to reflecting actual listeners' experience in terms of frequency response in those concert halls, not the "brighter" sound that can result from proximity and elevated microphone positions typical of many classical recordings, which over time may have resulted in changes to concert halls, i.e. the so-called "hi-fi concert hall."

Possibly the closest to the ideal design process might have been at the BBC, where direct comparison of in-studio and control room sound resulted in an interative design process of loudspeakers/monitors, microphones, and recording techniques?

Young-Ho

 

[Floyd Toole]

I tend to consider this as a cognitive illusion, since concertgoers can have different seating location and sonic preferences (see comments by Beranek and Lokki at https://www.audiosciencereview.com/...cert-hall-acoustics-links-and-excerpts.51487/), and virtually none of these seating locations correspond with typical recording microphone locations (https://www.psaudio.com/blogs/coppe...6XAzCGb2Lq2lf2zmJU7KXjcgK9NvfRHKNrYpUxWyU5Wuz), and the vast majority of us can only presuppose the engineers' intents without direct conversation or published description, even with which we must assume that description of intent completely and accurately captures the actual intent, while cognition typically involves at least two systems, one of which can occur below the level of awareness. With respect to the latter link, some criticized Kavi Alexander's Water Lily recordings of classical music as being too dull or muffled while others felt like it did come closer to reflecting actual listeners' experience in terms of frequency response in those concert halls, not the "brighter" sound that can result from proximity and elevated microphone positions typical of many classical recordings, which over time may have resulted in changes to concert halls, i.e. the so-called "hi-fi concert hall."

Possibly the closest to the ideal design process might have been at the BBC, where direct comparison of in-studio and control room sound resulted in an interative design process of loudspeakers/monitors, microphones, and recording techniques?

Young-Ho

I visited the BBC research department, toured around by Harwood himself, and saw how they approached the problem. They designed their BBC monitors for use in mostly smallish control rooms that were acoustically very well damped - i.e. primarily direct sound listening. this explains why some of their designs were not exemplary in off-axis frequency response. They also suffered from some licensed manufacturers that did not, or could not, maintain the written standard of performance. I have tested a few. Their goal seemed to be flattish on axis, which they came reasonably close to for the time. We can do better now.

 

[Arindal]

I visited the BBC research department, toured around by Harwood himself, and saw how they approached the problem.

Did you happen to visit them after 1999 again? If I recall it correctly, that was around the time when their approach changed, and they started collaborating with research institutions affiliated with other European broadcasting institutions.

Pls explain how additional speakers cannot possibly add more room modes or reflections?

The audible outcome of room modes does usually not depend on the number of them being excited (which will without any doubt increase with the number of loudspeakers), but rather how unevenly this is the case. The more speakers you bring in a room, the more likely you are to get mode excitement events cancelling each other out or at least leading to a more even distribution both in terms of positions and frequency bands.

I think it was Earl Geddes suggesting 4 or more subwoofers distributed over the room in order to cancel out modes. I tried that many years ago in a room of disadvantageous dimensions (2:2:1) and it worked surprisingly well, even better than a DBA under these specific conditions.

Similar thing with reflections: Even if you add a bunch by bringing in another loudspeaker, you increase the chance of them getting masked by direct sound in an overproportional manner due to more of angle being covered. This masking I would see as one of the main reasons why discrimination testing works better in mono or even better under anechoic conditions.

So, in the new advanced audio world, if I evaluate a truly neutral speaker like the studio used, in mono.... it will probably sound a bit bright.
Which will lead me to choose a duller non-neutral speaker, leading me to be disappointed when I move to stereo.

That would be my guess, maybe not even sounding brighter but adding proximity and subjective ´directness´, ´dryness´, hiding the reverb behind the direct sound. That was at least my impression after doing experiments with mono material played on stereo speakers.

The interesting thing in this case was, that it was not a dull speaker (tilted downward response under anechoic conditions), that was preferred, but a speaker with a linear on-axis response but dull off-axis response/increasing directivity index that created the effect that Dr. Toole has described as ´mono speaker disappearing behind the screen´.

I tend to consider this as a cognitive illusion, since concertgoers can have different seating location and sonic preferences

Okay, I recommend to sit there where the main mic is hanging, or a few rows closer to the stage.

Possibly the closest to the ideal design process might have been at the BBC, where direct comparison of in-studio and control room sound resulted in an interative design process of loudspeakers/monitors, microphones, and recording techniques?

I have been doing exactly such tests with recording engineers, musicians and critical listeners, and it turned out to be a failure. Recording engineers who read the score and mix in spot microphone signals according to their ideal of balance, clarity, proximity and localization, will always take the limitations of stereophony into account and ditch the idea of reproducing the whole reverb field in the concert hall. It always sounds different than ´the real thing´ and no-one can say what is better from the perspective of a critical listener. It is just different.

 

[Floyd Toole]

Did you happen to visit them after 1999 again? If I recall it correctly, that was around the time when their approach changed, and they started collaborating with research institutions affiliated with other European broadcasting institutions.

No later visits. Why would they change their approach, what to, and why? They were basically on the right track for their specific listening conditions. When these designs turned up in more reflective home listening rooms results varied, for good reason. The European broadcast industry standards for selecting loudspeakers and equalizing them are not completely rational, and they too specify very dead rooms: RT target 0.25s. They need updating. I have participated in a few international standards writing groups and they are not always populated by technically proficient people. An industry - i.e. money - is involved and there are "priorities", often related to not requiring changes.

 

[Arindal]

Why would they change their approach, what to, and why?

Towards control rooms with higher RT60, more even directivity of speakers, specific requirements for bigger control rooms.

The European broadcast industry standards for selecting loudspeakers and equalizing them are not completely rational, and they too specify very dead rooms: RT target 0.25s. They need updating.

That process of updating started in 1998, with partly the BBC involved, that was what I was referring to. We should remark that average RT60=0.25s was originally referenced for a small to medium-sized control room.

I remember the panels for standardization in this case to be surprisingly competent and open. External influence from any ´industry´ was barred out pretty quickly, which led to the unfortunate situation that almost nothing has been published.

 

[Heinrich]

... distance from the front and back wall with respect to length modes. ...

That was another question of mine, number three: in those mono listening sessions, with a single or multiple listeners in the room piling up on each other's lap, how was bass quality maintained? Reportedly it made 30% of the agreement.

I didn't refer to 'ideal' symmetric conditions, why should I? In my castle I have clearly asymmetric conditions stereo, and that helps - a lot! Conversely I expect from a listening room with only one speaker really bad results in that regard.

Btw: I would love to see the actual frequency response at seating position reported together with the results of tests in discussion. Not the spinorama or other in vitro data.

Did Harman tell? Good standard, we need it!

(If bass is harmanized by room treatment, a bias towards boomier speaker is introduced. Same with a too big room.)

 

[MattHooper]

Certainly true. But even if we accept that millions of recordings have been mixed and mastered on different systems, I would say there are a lot of similarities between studio control rooms, and the variations between most of them is much smaller than found between typical hi-fi setups at home. So if precise standardization cannot be achieved, we can still do some things to stay within the usual tolerance band of deviations found in studios. Listening in stereo and keeping and equilateral stereo triangle is the single most standardized aspect, while supressing disturbing early reflections achieving stable center phantom localization being the second most common. So why not follow at least these two?

Yes I have made similar arguments for that approach before. Even if you can’t re-create the listening scenario of all the mixing rooms you can in principal at least move closer or further away from that goal. For instance if you use a single loudspeaker for playback at home that would clearly be a move away from what they heard in the studio for stereo mixes. Placing your sound system in your tiled bathroom is also likely to be a move in the wrong direction, etc.

But of course, for all the reasons Toole has pointed out neutral playback at home does not avoid the circle of confusion. So one picks one’s compromises.

But as I argue, there are various legitimate roads to take based on any individuals taste or goals.

It can also make sense to take the stance many on ASR have voiced: at least as it stands, trying to re-create what they heard in all the different mixing theatres is a fools errand. That’s a total out of reach version of accuracy. Therefore I’m simply trying to be accurate to the recorded signal itself - my motivation is to reproduce it with as little distortion as possible.

You can end up with the same type of accurate and neutral system, but for different reasons.

Or you can be somebody who just simply doesn’t have either of those goals, but is simply trying to please himself. Many audiophiles, myself included, like to “ play” with sound. I’ve owned plenty of different loudspeakers, all of which sounded different - everything from Quads to MBL omnis, to egg shaped speakers to various takes on traditional box speakers. That they all sounded different wasn’t for me a bug it was a feature. I wanted to hear something different from different loudspeakers. That was the fun of it. And I still play with set up and acoustics in my room simply because I enjoy the different effects.

I don’t worry about things like “ the artist’s intent” mostly because I think any intent largely translates through a wide variety of sound systems.

And while I am not ultimately looking for the most coloured speaker I can find (I don’t want to feel anything is obviously missing in the frequency response and I do want to hear the character of different recordings) I’m also fine playing with colorations like vinyl playback, and my tube amplifiers.

Yes. But how do you define which of your favorite tracks gives you the most of enjoyment from sonic point of view?

I decide this based on how the majority of my music sounds on my system not just a few tracks.

If your system deviates from the average studio tolerance band, one recording might sound fabulous, the next one annoyingly kinked.

But the same could be said for a perfectly neutral system. Due to the variable quality of recordings you’re going to get variable sound quality with playback on a neutral system as well. So the neutral system doesn’t in principle solve that problem. There’s a circle of confusion again.

Since your point there has to do with what somebody finds pleasant or not, if somebody finds a coloration they happen to like across a broad spectrum of their music (and it doesn’t have to be a gross colouration), then that can be just as satisfying - perhaps even more satisfying for them - then having chosen a perfectly neutral system.

To leave the circle of confusion, I plea for taking the percentage of recordings which ´just sound right´ and give enjoyment if they are within the genre limits of the listener´s taste, as measure how well the average sound reproduction is met.

I understand the argument you are making there - the more recordings that sound good on a system the more likely your system is getting closer to what they heard in the mixing studio. That doesn’t escape the circle of confusion really but still there might be something to that.

Be that as it may, as has been pointed out already, it seems audiophiles have found satisfaction with quite a variety of speaker, designs, and performance (and we can adapt to colorations). Over the years I’ve seen the occasional person moving on from a high quality Revel speaker to a more coloured speaker because they felt not enough of the music they listen to sounded pleasing on the Revel speaker, and they found more happiness with another speaker.

If I were making recommendations for a newbie, I would first be recommending well design loudspeakers like Revel or other affordable well measuring speakers. But somebody still may end up cutting their own path.

On the other hand, something between 80-90% of degree of satisfaction is achievable.

For me, I find having taken the “ set up a system that I find pleasing” approach, and even introducing a little bit of colouration, has yielded without exaggeration something more like 99% satisfaction rate with what I’m hearing. And that is with a collection of music of wildly divergent production quality. I’m perfectly happy with the variation between recordings as I find that interesting in of itself, and then I’ve introduced just the type of colouration that I find pleasing across virtually all recordings.

The other method of leaving the circle of confusion is visiting live concerts without sound reinforcement. From my personal point, the single most reliable way of judging if a system in a room ´sounds right´ is to visit a concert which will be subject to a broadcast or album issue later on. Knowing both, preferably with little delay between both listening sessions, gives you an astonishingly good understanding of how things should sound and what the mixing engineer was meaning. I personally prefer to hear what the broadcast mixing engineer is doing during general rehearsal, before I visit a concert, but I understand that is not doable for everyone.

That’s fun stuff! And I’ve been in bands where we’ve recorded in the studio, as well as doing my own recordings for live versus reproduced tests. I think this type of experience can be interesting.

But I certainly wouldn’t see how it would solve the circle of confusion which extends far beyond the limited experience one can have of the type you’re describing.

Cheers

 

[Floyd Toole]

Towards control rooms with higher RT60, more even directivity of speakers, specific requirements for bigger control rooms.



That process of updating started in 1998, with partly the BBC involved, that was what I was referring to. We should remark that average RT60=0.25s was originally referenced for a small to medium-sized control room.

I remember the panels for standardization in this case to be surprisingly competent and open. External influence from any ´industry´ was barred out pretty quickly, which led to the unfortunate situation that almost nothing has been published.

I am currently proof reading the text edited version of the 4th edition - digressing for "relaxation" on ASR Forum I came across this comment I made about two well-referenced standards and wondered what you thought:

ITU and EBU recommendations. The ITU-R BS.1116–3 (2015), EBU Tech 3276–2nd edition (1998), and EBU Tech 3276-E (2004) documents are often referenced for guidance in setting up systems for sound reproduction. The loudspeaker and room acoustical requirements are identical, yet they have very different purposes.
• The ITU document is intended for blind listening tests evaluating codecs having “impairments so small as to be undetectable without rigorous control of the experimental conditions and appropriate statistical analysis.” It is meticulously technical in its objectives, asking the question: is there a perceptible difference between two sounds, and how large is it? This is done in rigorously administered blind tests using listeners who are audiometrically tested, trained and statistically evaluated.
• The EBU documents are intended “for the critical assessment and selection of programme material for inclusion in a sound or television broadcaster’s programme output.” In distinct contrast, the application of the EBU document is artistically subjective, asking the question of recording engineers: does the program sound good enough to be broadcast?
The ITU listeners look for differences between two sounds, relative judgments. The details of the listening circumstances are constant factors, and placing the listeners in a dominant direct sound field may be advantageous. The EBU listeners are making absolute judgments of sound and spatial qualities. The performances of the loudspeakers and room are inseparable parts of the sound they are judging. These listening experiences become the reference for the “translation” of the art. It is the “Creation of the Art” component in the circle of confusion shown in Figure 1.8.
Audio science has progressed, and these recommendations are not up to date. Academics, particularly, rely on standards to validate their results, so it is important that the recommendations are themselves validated.

 

[Travis]

@Floyd Toole

First, Thank you for your time and drilling down more into some of these issues and questions from your decades of research. I’m very much looking forward to the 4th Edition of your book which I understand is going to have an updated and expanded section on recording studios.

I think this question directly relates to the original post, and gist of the thread. What is your recommendation for the consumer who believes in the science, and is purchasing floor standing speakers for their large American living room. They look at the Spinorama data (FR and DI), may even consult predicted Preference Score, w/wo sub, find 3 Brand/models of floor standing speakers that look to fit budget and other parameters, what is your suggestion on how they should evaluate these speakers in their home?*

I thought I had an idea on how to go about this and then caught your recent interview with Erin and you reminded me that “auditioning them for a week” to see how they sound is fruitless because of acclimation. [Edit: after a re-listen of the interview what you actually said was “adaptation sets in.”] Had to go back to your book and remind myself of that.

It seems to me that what has to happen is you have to get a really good friend who can set Model A and Model B, initially, up in your living room without you being in the room, and assuming they can level match them properly with an A/B (left right) switch, then leads you in blindfolded to the listening chair, plays the pink noise, plays the obligatory “reference” tracks including the 1st 30 seconds (but no more) of Ms, Chapman’s Fast Car and you say, “out of those two, I would go with ____” and maybe also state by how much so. “Very close but slight edge to L, or R by a country mile.” Then get led outta the room with blindfold, so audio buddy can remove the one (possibly the one that won in round 1 just to keep you on your toes) and bring in Model C for another A/B, and in the interest of science, do a 3rd round so all 3 have been compared to each other at least once. Hopefully the selection was consistent.

Let’s even assume all 3 have very similar bass response so we don’t need to account for the 30% factor for bass performance.

Back to this thread, let’s say there is clear hierarchy of 1, 2 and 3, is a repeat blind listen of maybe 1 and 2 but now in stereo even worth while (assuming it takes 10 mins to change from 1 to 2)?

What do you suggest for the consumer, who believes in the science, has researched available Spinorama/preference scores, but before making a $10k - $20K (Eg. M2) purchasing decision they want to (blasphemy I know) give them a little listen.

There is constant stumbling in the Forum when members post a question like “Model A, B or C for my Living Room.” They will offer go to great trouble to give dimensions and photos of room, primary use (HT, music only). The responses are, for the most part, well thought out and intentioned, but run the spectrum of what sounded best to them, they own Brand B and can’t say enough, what’s in the Spinorama graph data they like/don’t like, or quoting you on sighted listening evaluation.

What do you suggest for the consumer, trying to decide on floor standers for their large living/music room, in making their final selection with a listening test?

*I picked floor standing speakers because of the 86% preference score correlation and a listen at home from your statement in your Edit “Audio-Science in The Service of Art” paper https://www.harman.com/documents/audioscience_0.pdf that we still need to listen. I think it was in the context of design and manufacture but assume it also applies to the end user.

 

[Arindal]

I came across this comment I made about two well-referenced standards and wondered what you thought:

The way you pointed out the difference between discrimination test and overall preference test (overall in the sense of judging the sound of the broadcast stream, loudspeakers and room as a whole entity), is accurate.

My personal impression would be that the idea formulated by the ITU is not of such relevance anymore. Particularly for discrimination tests involving lossy codecs, I would assume using headphones has become a more popular method.

the application of the EBU document is artistically subjective, asking the question of recording engineers: does the program sound good enough to be broadcast?

Maybe this part is worth further explanation, as ´sounds good enough to be broadcast´ is maybe not a precise explanation of the reasoning. In my understanding, the idea of standardizing control room listening conditions was made by the EBU from the point of interoperability as seen from the perspective of balance engineers and recording engineers in any member institution, any country and any facility. You can loosely compare it to the importance of monitor calibration for people producing and subjectively judging video content from a technical perspective.

In practical terms, a recording engineer could be listening to the mix of a live broadcast from a Spanish broadcast truck, a British editing suite or a German master broadcast control room and would always come to the very same conclusions whether the mix should be altered according to his subjective judgement or not. When there had been no standardization, it was not uncommon that

- in the truck parked in front of the church where the performance takes place, a massive 50Hz humming noise in the signal would go on air unnoticed due to overly compact speakers incapable of reaching that low
- in the editing suite the engineer would be tempted to add reverb from the ambience microphones and broaden the stereo base due to non-standard stereo triangle
- in the master control room he or she would feed the necessity to reduce bass because the mix sounds bloated and boomy on the big system

This example might explain why certain aspects like stereo angles, listening positions, RT60 and SPL of the noisefloor were paid such amount of attention to. It was never about performing controlled listening tests but reducing tolerances induced by room and loudspeakers to a level the mix can be judged everywhere.

AFAIK the last update was when control rooms have been equipped with 5.1 capability in masses. I agree that the recommendations are not sufficient to ensure the initial goal of standardization. Discussing the reasons would be subject for a separated thread. Nevertheless they have created awareness for the important aspects of the listening conditions in control rooms, and situation has massively improved over the course of the last 27 years since the first group had published their recommendations.

 

[MattHooper]

A long audition on a single model in your home provides almost no useful information

Overall what I took away is: if you have ability to do a blind 3 or more speaker comparison you can come away with some useful information. Short of that it’s a complete guess.

This type of language strikes me as exaggeration.

Admittedly, I have typically had more than one loudspeaker in my home to compare (sometimes up to seven different pairs) but even the idea that just having one speaker would yield no use useful information seems highly exaggerated.

From the day I heard my current loudspeakers in the store (before encountering any reviews ) to six years later of using them in my home, they have sounded essentially the same to me, and the way they sound to me matches the Stereophile Measurements quite well: very clear and clean sounding, excellent “ disappearing” as an apparent sound source, capable of wide deep soundstaging with precise imaging, with a little bit of added richness in the bass, fairly flat midrange, a little bit of elevation in the upper treble that gives a nice bit of sparkle and airiness but without causing undo emphasis in vocal sibilance, and nice off axis performance so that the sound seems fairly consistent when I move around the room.

I heard all of those characteristics from the moment I started listening to those speakers, well before I’d seen any measurements, and nothing at all has changed about my impressions. An audio friend who came over and listened to them described exactly the same qualities just from listening to them, not aware of any measurements at all.

In fact, over and over I have found that my listening impressions have tended to match quite well to measurements for a loudspeaker. If I detect a sway back in the midrange, it’s there in the measurements. If I detect a lack of mid or low bass it’s there in the measurements. If I detect a scoop out in the warmth region of the midrange or some emphasized or ragged high frequencies… usually it’s there in the measurements. (and I also get to hear lots of loudspeakers at my friends place before they are sent to the NRC for measurements afterwards. It’s not unusual for us to have identified characteristics that show up in the measurements - for instance my friend was bothered by a big trough in the midrange of a recent loudspeaker, and that showed up in the measurements later on).

I am in no way a particularly talented listener. None of this is to dismiss the obvious problems of bias that are well documented. I’m just saying that the type of all or nothing assessments I sometimes see “either you are comparing several loudspeakers under blind conditions” OR you are just totally guessing at the sound and your impressions are inaccurate … that just strikes me as exaggerated and does not explain either the consistency of my impressions of loudspeakers I own, or the general consistency between what I perceive when listening to different loudspeakers and what shows up objectively in their measurements. Or for that matter as I brought up before, the apparent consistency between Erin’s subjective impressions and descriptions of loudspeakers and how they often comport with the measurements.
(and again… if the measurements had no predictive quality to what we are going to be able to perceive, then they would be useless).

 

[amirm]

Unfortunately, Erin never asks Dr. Toole “so knowing all of this, how can the consumer incorporate all of this great research in a buying a pair of speakers? Just assume they have the ability to get 3 or 4 pairs of speakers in their home at the same time, a really great spouse/audio friend who is willing to help (in the interest of science), how should they go about trying to make the best choice possible for their home?”

I have made the above point many times when people demand that I "listen before I measure." And they go on to say so and so does it that way. I keep saying that such evaluations are essentially random but people rely on lay intuition that they must be valid. Never mind that they are sighted and subject to failures you quoted from Dr. Toole.

My solution to problem is twofold:

1. Rely on measurements and have it be that. Don't try to second guess them by listening to sighted evaluation of some reviewer. Your worse case risk is quite low and essentially zero if you have the ability to EQ to taste.

2. If Measurements are not perfect, I create PEQ filters and test each one blindly if needed. Then, I have the type of comparison Dr. Toole mentions. It is remarkable how the picture changes/understanding improves of the impact of frequency response errors in this type of instant AB testing. I will go as far as saying this is more sensitive testing than the speaker shuffler as that takes in order of 4 to 5 seconds to switch speakers. When speakers are tonally close, that is way too long to hold on to memory of their tonalities. In contrast, I can do AB switching in under a second (as fast Roon can switch the filter on/off).

While I also make general fidelity comments, the most accurate and reliable part of my "listening tests" is application of EQ.

Net, net, I completely ignore sighted, single speaker tests from anyone. Often I find that they are cheating and using measurements to know "there is a peak around 400 Hz" even though they claim to have done listening tests before measurements.

Anyone can do #2 once they have the measurements. If you are buying speakers without measurements, then you are lost forever. There is no hope really. Even I would be guessing and could be wildly wrong. Measurements are our compass and potentially GPS in a thick jungle.

 

[Travis]

There is no research that supports the claim that adaptation eventually wears out.

Yes there is, but it’s in the field of Neuroscience. Dr. Howard Nusbaum has been studying frequency and music’s effects on the brain for sometime. His lab is equipped for high-density EEG measurements, fMRI data analysis (with access to a 3T scanner), and auditory brainstem recordings.

There is a study from about 10 years ago where he studied subjects with absolute ( “perfect”) pitch who “adapted” to a song that was slowly flattened by a third of a note (33 cents) in the first part of the song over several minutes. It was done so slow they didn't perceive this was happening. When the song was over they were replayed the flattened song and the early part where the notes were correct they identified as out of tune. When it got to the part that was fully flattened (out of tune by 33 cents) they identified them as the correct notes. They had adapted within minutes while the song was played, and remained adapted when the modified tune was played back to them.

Later they sat at a piano, played notes on a properly tuned piano and they adapted right back to their pre-experiment absolute pitch ability.

 

[amirm]

There is no research that supports the claim that adaptation eventually wears out.

That was a compound sentence. The wear out part related to you saying "visual impression." That sighted placebo effect goes away as placebos do. Otherwise, we would all be taking placebos instead of real medication.

I addressed the adaptation separately, explaining that you will realize what you are hearing is not right when you listen to a neutral system. Then go home and immediately realize yours is too bright, mid-range sucked out, etc.

Finally, how do you know any research doesn't exist? You have studied every paper? What you don't know is neither here, nor there. But sure, go ahead and show how sighted bias lasts forever.

 

[Travis]

My hypothetical 3 speakers and how to decide on a final choice while adhering to the science has a lot built into it.

1. The speakers all measure well per Spinorama. I wanted to leave predicted preference score out of it, but let’s assume they are within the -/+ limits of the score.

2. They all have a very similar bass response because of the 30% weighting of bass response in predicted preference score.

3. They have someone who can “level match” them at the listening position (which is very difficult to do). The A/B part is easy.

4. They have a blindfold that works

Given @amirm comments in his post above about EQ, let’s assume all three speakers are also within the margins of error for the preference score as a raw score, and likewise within the margins of one another on the “with eq” preference score. (I think @Sean Olive Dr. Olive said some where this was “1”. A 6.2 and a 7.2 are going to be close enough).

This, per the science, will result in very close cases, or even in statistical ties among the 3 pairs of speakers in controlled testing environment (blind, auto shuffler) with an 86% correlation for these hypothetical floor standing speakers.

If a consumer wants to satisfy herself on what will sound best in her music room, how should this consumer go about doing this?

I suspect, different stakeholders would want a different answer from Dr. Toole. The manufacturer, if they have a good measuring speaker, would want the answer to be: “Per the CTA Standard, it’s “GOOD” that’s all you need. Have your dealer deliver and set up, listen to confirm, the score takes the guesswork out of it.

The dealer is going to want the answer to be “go to your dealer, have a listen, if not convinced they will set up a blind A/B(/C/D) for you in their typical music room then you take them home, we set up, you give a final okay.”

Then the reviewers will want a different answer, based on whether they have a NFS, and/or RTA.

What does the consumer do? It could be one of those situations where the scientists/researchers said “we never considered that, this was for speaker designers, to come up with measurements that would provide a high degree of predictability if they achieve these measurements, they will have a good sounding speaker.” (Unlike the Consumer Union testing at the time that almost had an inverse relationship-the better the score the worse the measurements). Maybe it’s invalid trying to carry the comparisons and final selections into the home?

“Whatever you do, please don’t try this at home.”

 

[Scytales]

I was wondering what happened to the one chair that was missing when I was there!!!

I am sorry to derail from the topic of this thread, but I would be interested to read your comments about the effect of the white diffusors.

To my eyes, the total surface of the diffusors seems very low and I wonder about the benefit they procure and why they have been placed where they are.

 

[Floyd Toole]

Yes there is, but it’s in the field of Neuroscience. Dr. Howard Nusbaum has been studying frequency and music’s effects on the brain for sometime. His lab is equipped for high-density EEG measurements, fMRI data analysis (with access to a 3T scanner), and auditory brainstem recordings.

There is a study from about 10 years ago where he studied subjects with absolute ( “perfect”) pitch who “adapted” to a song that was slowly flattened by a third of a note (33 cents) in the first part of the song over several minutes. It was done so slow they didn't perceive this was happening. When the song was over they were replayed the flattened song and the early part where the notes were correct they identified as out of tune. When it got to the part that was fully flattened (out of tune by 33 cents) they identified them as the correct notes. They had adapted within minutes while the song was played, and remained adapted when the modified tune was played back to them.

Later they sat at a piano, played notes on a properly tuned piano and they adapted right back to their pre-experiment absolute pitch ability.

Fascinating. In my PhD work in the early 60s, I did some experiments correlating visual and auditory positional localization using an in-head (headphone presented) sound image and a visual "blob" on a TV screen. The experiment started with the images in sync, but I gradually introduced a time delay bias in the auditory image. The listener image tracking responses showed that the visual cue was dominant - they did not notice the discrepancy. After about an hour a purely auditory localization test was done and the results showed that they had adapted: zero interaural time difference (ITD) was no longer judged to be a centre image. Thus was disproved a longstanding psychoacoustic belief in sound image "centering" experiments that relied on the notion that zero ITD and centre localization was a "hard wired" property of humans. The adaptation faded quickly once listeners were exposed to the real world of correlated visual and auditory cues. This experience never left me. I have since never ignored the reality and power of adaptation. Your example is more proof. Thanks.