Which speakers are the Classical Music Pros using?

[JJB70]

Much to my amazement, I've noticed a lot of musicians don't care. They just ask for a regular one, no fuss. They often tell me any which one will do, just as long as its regular and does the job. They seem to listen to some other aspects. I've seen pro classical musicians with anything from small Yamaha bookshelves to whatever is available at the shopping mall. Also, very often I see they get some of the equipment through whatever club/society/foundation funds the orchestra and in that case it's the ones locally manufactured. In Vienna, I've seen a lot of musicians having Vienna Acoustics. In Italy, it's Sonus Faber (smaller and cheaper ones)

This doesn't surprise me, most musicians I have met have been uninterested in audio gear, audio gear generally doesn't matter to them.

 

[Floyd Toole]

Thank you for your replies.



I understand the need for blind testing as a means to remove bias.
I have found that familiarity provides the highest level of discrimination in a listening assessment, and that is achieved over long term listening with familiar music and in a familiar system/room, and in that way assessing only a single variable/change to the system/sound.
Is this something that you considered when developping the methodology for your tests and testing facilities?



I have stressed the importance of familiarity but another extremely important factor in my view is the adequate set up of speakers in the room for best bass frequency response (and your research indicates that "low bass performance accounts for approximately 30% of one’s overall assessment of sound quality") and room "activation" (how differences in directivity interact with the room for "spatiality" effects, such as dipoles and omnis), and axial positioning or toe-in (e.g. Dali speakers are designe for flat response 30° off-axis).

Would you confirm that neither aspect was/is addressed in the shuffler rooms?
If that was not the case then some speakers were listened in sub-optimal conditions and thus unfairly handicaped.
Although I understand that providing optimal conditions for testing would have been impractical.

In regard to the mono vs stereo performance, your research shows that "spatial quality" ratings when listening in stereo improve for narrowing directivity monopoles (Kef) and for dipoles (Quad).
This seems to indicate that wider directivity is less important in stereo pairs, which is how speakers have been used for 50 years, and thus my interpretation of the data would have led me to dismiss mono as a means to assess "spatial quality".

Also, I would expect that anomalies in both the axial response and also the quality and amount of bass would still have influenced the listener preference when assessing "spatial quality" (as you've mentioned, "if something doesn’t sound timbrally correct I don’t much care about space"). Were these aspects corrected through optimal positioning and/or high-passing and the use of EQ?
I understand that again the question of practicality arises. It would be difficult to find even a pair of prototypes of the same speaker one with wide- and the other with narrowing-directivity, but that in my view would have prevented the introduction of other variables which undermine the effectiveness of the tests. Once we introduce more than one variable we no longer know for certain what the listeners are reacting to (is it the different bass response, a dip in the presence region, or how the speakers interact with the boundaries is generating a more pleasing level of "envelopment" and "spaciousness"?).

For the reasons highlighted I feel reluctant to agree with your interpretation of both the adequacy of mono testing for "spatial quality" assessment and, consequently, of the results which resulted from that testing.



Harman's "target curve" research also seems to indicate that untrained listeners prefer a lot more bass and more treble (sloping upwards with frequency). This is also my impression from observing people's reports on different speakers and show systems.
Isn't this indicative that people have different preferences when it comes to tonal balance?

I think most would agree that the ultimate goal of a playback system is to provide listening enjoyment to the end user. And because we have different tastes in music and in "presentation", and different rooms, perhaps creating a standard on-size-fits-all kind of speaker is a disservice to the community.






There's no doubt that the research you conducted was pioneering and produced a significant amount of valuable data. But as mentioned earlier I have some reservations in regard to both some of the metodology used in the tests as well as the interpretation of some of the data, but I am just a curious and inexperienced amateur with a lot of questions.

(I wish I could have expressed myself more clearly and eloquently but my means of expression is the drawing not the word, and English is not my first language)

Tuga said: “I understand the need for blind testing as a means to remove bias.
I have found that familiarity provides the highest level of discrimination in a listening assessment, and that is achieved over long term listening with familiar music and in a familiar system/room, using only a single variable. Is this something that you considered?”

Of course it was considered. It is called the “single stimulus” method of evaluation, which should also be done blind to avoid bias. The “take it home and listen to it” style of evaluation is a default, done in the absence of anything better. It is the norm among subjective reviewers. Comparison tests are generally more revealing of differences, but even then, if there are only two products and they share a defect, it may go unnoticed.. We found that comparisons among 3 or 4 randomly presented sounds (loudspeakers) was extremely revealing of differences that went unnoticed in prolonged exposure to single sounds, where adaptation (a profound capability of humans) is a major factor. We learn to “listen through” many kinds of technical peculiarities and flaws to be able to enjoy the music. What you are describing as your preferred method is not blind (and therefore subject to bias) and generously allows for adaptation. Familarity with the music is not necessary, only the ability to discern aspects of its reproduction that are not natural or pleasing. An important aspect of listening blind to several versions of the same program is that one quickly identifies the timbral features associated with the individual loudspeakers from those that are “constant”, namely the program itself and the room. Not all programs are equally revealing of differences: Olive, S.E. (1994). “A Method for Training Listeners and Selecting Program Material for Listening Tests”, Audio Eng. Soc. 97th Convention, preprint 3893.

Over the many years of doing these evaluations we have encountered numerous people who shared your view – including virtually all money-earning subjective reviewers. When they experienced the double-blind multiple-comparison test they performed no better than “ordinary” people: Olive, S.E. (2003). “Difference in Performance and Preference of Trained versus Untrained Listeners in Loudspeaker Tests: A Case Study”, J. Audio Eng. Soc., 51, pp. 806-825.

The reviewers often commented that they wished they had the facilities we had, but lacking them, did what they were able to do and felt comfortable with. You seem to be in that camp when you say: “familiarity provides the highest level of discrimination”. How do you know?

BTW, people who think they have “tin ears” usually turn out to perform normally. Those listeners who are distinctive or indecisive in their opinions usually have hearing loss. Musicians do not appear to have advantages, in fact many rationalize – I recall one saying about a mediocre loudspeaker that: “it is a valid interpretation of a cello”. He was listening to the music.


Tuga said: “In regard to the mono vs stereo performance, your research shows that "spatial quality" ratings when listening in stereo improve for narrowing directivity monopoles (Kef) and for dipoles (Quad).”

Here is a Figure 7.14 summarizing that research:

It shows that in terms of sound and spatial quality ratings, mono evaluations were much more revealing of differences. The additional binaural effects in stereo recordings were perceptually rewarding to be sure, but in ways that disguised timbral differences between loudspeakers. The recordings themselves were the dominant factors. Spatial Effects in mono reproduction? Yes, that surprised us, but when listening it was clear that the best, most neutral, loudspeakers came closest to “disappearing” behind the visually opaque screen, revealing depth information in recordings. Further analysis revealed that the mono pattern of ratings was closely replicated in multi-mike pan-potted popular recordings in which hard-panned sounds are reproduced by single L or R loudspeakers. Classical recordings were quite inconsistent.

What were you referring to?

“Were these aspects corrected through optimal positioning and/or high-passing and the use of EQ?”

Of course we did not play with filters or EQ – we were evaluating speakers as they were manufactured. I would not know what “optimal positioning” is for a loudspeaker of unusual or aberrant design unless the manufacturer specified it. If it was specified, it was obeyed – e.g. Allison boundary-friendly designs. At Harman we were most interested in evaluating competing products. We did not have a budget to purchase off-beat, small-distribution products, whatever possible virtue they may have had. Still the list of products evaluated in detail is long.


“Harman's "target curve" research also seems to indicate that untrained listeners prefer a lot more bass and more treble (sloping upwards with frequency). This is also my impression from observing people's reports on different speakers and show systems.
Isn't this indicative that people have different preferences when it comes to tonal balance?”


Of course people can have preferences in spectral balance. This is why I poke fun at “High End” products that don’t have tone controls. Why not let customers buy broadband, neutral, loudspeakers and let them compensate for program deficiencies or indulge personal preferences, always with the ability to return to neutral. In some of the tests you refer to loudness was not compensated for, and it was sometimes thought that the bass and treble boosts used by younger, inexperienced, listeners could just have been a way to turn the volume up. That said, it is obvious that some categories of contemporary music thrive on exaggerated bass.

“ I am just a curious and inexperienced amateur with a lot of questions.”

Curiosity is a good thing – it is what drives research. I too was once an amateur, but rose above that station by applying the scientific method to my activities. The world and internet forums are filled with amateurs with opinions formed under circumstances where bias and adaptation are possibly as important contributing factors as the physical realities. If you are unable to apply the scientific method, the next best approach is to study that which as been done - which is the process you are in right now. So, I will terminate this personal tutorial and let you get on with reading more of the science. Keep an open mind, and enjoy.

 

[Mauro]

In other forums on this site I have commented extensively on the value and meaning of room measurements and "correction". I published a paper, which you can freely access that explains some of it: Toole, F. E. (2015). “The Measurement and Calibration of Sound Reproducing Systems”, J. Audio Eng. Soc., vol. 63, pp.512-541. This is an open-access paper available to non-members at www.aes.org http://www.aes.org/e-lib/browse.cfm?elib=17839. "House curve" is a term used in pro audio in days when steady-state measurements were all that could be done and it was assumed that each venue had an "ideal" curve. It was wrong but the term still lingers. The direct sound is what matters most as a starting point in venues of any size. Once the sound has been launched into a reflective space - any normal room - it is not possible to interrogate the detailed performance of a loudspeaker from measurements at the listening location. Comprehensive anechoic data are essential to understanding what a "room curve" is showing. This is usually absent, so there develop endless discussions of "optimum" room curves.
In brief, I used REW, a free downloadable measurement program, and a single calibrated mic located at 2 m on axis to evaluate and correct the direct sound - which the "automatic calibration" had seriously altered. This is an essential factor to get right. This was confirmed at the listening location with a small spatial average to assess the bass performance and to finess the EQ of the four subwoofer "Sound Field Management" scheme I used (described in Section 8.2.8 in the 3rd edition).

Is that Chapter in the 3rd edition covering the same content of the 13th Chapter in the 2nd edition (which I have)?

I would appreciate so much having from you a guideline to properly measure and apply eq to edit the automatic correction suggested curve..

Currently that’s what I used suggested by an ex jbl engineer, now Kali Audio’s. Any comment on this? (thanks @Charles Sprinkle btw

)
In case someone might be curious, here you can find my L+R averaged measurements in green overlapped as best as I could with Sean Olive home theater’s curve used for reference.

 

[Floyd Toole]

Is that Chapter in the 3rd edition covering the same content of the 13th Chapter in the 2nd edition (which I have)?

I would appreciate so much having from you a guideline to properly measure and apply eq to edit the automatic correction suggested curve..

Currently that’s what I used suggested by an ex jbl engineer, now Kali Audio’s. Any comment on this? (thanks @Charles Sprinkle btw

)
In case someone might be curious, here you can find my L+R averaged measurements in green overlapped as best as I could with Sean Olive home theater’s curve used for reference.

"Is that Chapter in the 3rd edition covering the same content of the 13th Chapter in the 2nd edition (which I have)?"
There is overlap, but the 3rd edition is significantly expanded and improved. It is an almost complete rewrite of the original book.

 

[Mauro]

"Is that Chapter in the 3rd edition covering the same content of the 13th Chapter in the 2nd edition (which I have)?"
There is overlap, but the 3rd edition is significantly expanded and improved. It is an almost complete rewrite of the original book.

That’s a shot to my heart! I bought the second edition a couple of months before the third edition came out! Sad story..

 

[Duke]

That’s a shot to my heart! I bought the second edition a couple of months before the third edition came out! Sad story..

I have both versions as well (the content of the first and second are identical). But look at it this way - now you have EVEN MORE great stuff to look forward to. For example, section 7.4.6 is Floyd Toole describing his personal experiences as an audio enthusiast. Imo the third edition is WELL WORTH the price of admission.

 

[tuga]

Tuga said: “In regard to the mono vs stereo performance, your research shows that "spatial quality" ratings when listening in stereo improve for narrowing directivity monopoles (Kef) and for dipoles (Quad).”

Here is a Figure 7.14 summarizing that research:

It shows that in terms of sound and spatial quality ratings, mono evaluations were much more revealing of differences. The additional binaural effects in stereo recordings were perceptually rewarding to be sure, but in ways that disguised timbral differences between loudspeakers. The recordings themselves were the dominant factors. Spatial Effects in mono reproduction? Yes, that surprised us, but when listening it was clear that the best, most neutral, loudspeakers came closest to “disappearing” behind the visually opaque screen, revealing depth information in recordings. Further analysis revealed that the mono pattern of ratings was closely replicated in multi-mike pan-potted popular recordings in which hard-panned sounds are reproduced by single L or R loudspeakers. Classical recordings were quite inconsistent.

What were you referring to?

Looking at that figure I agree that mono evaluations are more revealing of differences.
And perhaps one can conclude that wide directivity rates higher than narrow or narrowing directivity in mono both in terms of spatial quality as well as sound quality.
But when the speakers are evaluated in stereo (their intended use) the wider directivity speakers rate just as highly as the others in both parameters.
Whilst I agree that mono evaluation is the more adequate method for assessing the sound quality of a single speaker, perhaps assessing spatial quality in mono is only revealing of how well a speaker performs by itself, and speakers are meant to be used in stereo pairs in which case the "advantage" of wide-directivity vanishes.

“Were these aspects corrected through optimal positioning and/or high-passing and the use of EQ?”

Of course we did not play with filters or EQ – we were evaluating speakers as they were manufactured. I would not know what “optimal positioning” is for a loudspeaker of unusual or aberrant design unless the manufacturer specified it. If it was specified, it was obeyed – e.g. Allison boundary-friendly designs. At Harman we were most interested in evaluating competing products. We did not have a budget to purchase off-beat, small-distribution products, whatever possible virtue they may have had. Still the list of products evaluated in detail is long.

If the speakers were not positioned in their optimal position (using the manufacturer's indications and informed by measurements) then soem speakers were at an unfair disadvantage. And looking at a few photos of the shufflers it doesn't look like care was taken in that regard. And the same is true for not EQ'ing the response of the speakers; not doing so conflates two variables – tonal balance and directivity characteristics –, the “single stimulus” method of evaluation was not complied with.

 

[youngho]

That’s a shot to my heart! I bought the second edition a couple of months before the third edition came out! Sad story..

The second edition still has a lot of useful information that does not appear in the third edition, so I often refer back to it.

 

[Floyd Toole]

The second edition still has a lot of useful information that does not appear in the third edition, so I often refer back to it.

Yes, being limited to 500 pages was a serious constraint, forcing difficult choices about what to discuss. The original book has more "hard science" discussion.

 

[youngho]

Yes, being limited to 500 pages was a serious constraint, forcing difficult choices about what to discuss. The original book has more "hard science" discussion.

I appreciated the online supplementary material for the third edition, but I I found the data and measurements in the original to be invaluable.

While I might have your attention, could you possibly comment on your opinion on the relative importance of "relatively constant" versus "smoothly changing" directivity, as well as the frequency ranges where this is crucial? Many of the Harman speakers (JBL LSR 6332 or Revel 228Be, for example, though the Revel Salon 2 comes relatively close) have what I consider to be a stair-step directivity index with relatively constant directivity from several hundred hertz to several thousand hertz, with rising directivity above 8-10 kHz. Others have a directivity index that more closely approximates a diagonal line (which I consider "smoothly changing"), like some of the JBL HDI speakers or possibly the JBL Array 1400, which I believe is speaker B in Figure 18.17 of your original book, which is smoothly changing up to about 3 kHz and then relatively constant above that, but others like many Kef or Genelec speakers have a more gradual semi-diagonal line.

My speculation is that listeners especially sensitive to tonality or timber may possibly prefer speakers with the "relatively constant" directivity index because the off-axis response has more spectral similarity to the direct signal in a critical frequency range (several hundred to several thousand hertz, also taking into account Gresinger's comment regarding instrument timbre 2-4 kHz). On the other hand, speakers with steadily rising directivity would seem to be expected to have increasingly dissimilar off-axis spectral content relative to frequency.

Furthermore, I wonder whether it's also possibly that more highly furnished rooms with "relatively constant" directivity speakers end up having a relatively similar room curve with more Scandinavian or spartan rooms with "smoothly changing"ones. My informal hypothesis is that different listener preferences should be considered in loudspeaker choices/recommendations (and modifications to the listening environment), hence preferences for reverberation=extremely wide dispersion like omni/bipolar (and more scattering/diffusion), loudness/proximity=wide dispersion, definition/clarity=narrow dispersion (so more absorption of first reflections, also with a special case of dipolar for listeners divided between definition/clarity and reverberation), timbre=relatively constant DI over critical frequency range, and bass extension.

 

[youngho]

But when the speakers are evaluated in stereo (their intended use) the wider directivity speakers rate just as highly as the others in both parameters.
Whilst I agree that mono evaluation is the more adequate method for assessing the sound quality of a single speaker, perhaps assessing spatial quality in mono is only revealing of how well a speaker performs by itself, and speakers are meant to be used in stereo pairs in which case the "advantage" of wide-directivity vanishes.

If the speakers were not positioned in their optimal position (using the manufacturer's indications and informed by measurements) then the “single stimulus” method of evaluation was not complied with. And looking at a few photos of the shufflers it doesn't look like care was taken in that regard. And the same is true for not EQ'ing the response of the speakers; not doing so conflates two variables: tonal balance and directivity characteristics.

I don't think that you've read Toole's replies or considered the concept of testing a single variable at a time carefully enough. Otherwise, one might look at the information presented already and conclude that nothing really matters in stereo, anyway, since everything rates relatively similarly. Similarly, one might actually believe that two pairs of speakers placed at different locations (is such a thing as "optimal position" even possible to define? What about toe-in?) in the room and equalized (how? Direct on-axis or at the listening position, what kinds of filters, effects on phase or delay, etc) can be compared in a valid way scientifically, as if multiple variables were not being conflated therein.

 

[Newman]

I agree. The fact that stereo listening lifted the spatial scores of speakers that performed worse on that metric in mono, is not a logical reason to say stereo is better for judging spatial ability.

 

[Newman]

Yes, being limited to 500 pages was a serious constraint, forcing difficult choices about what to discuss. The original book has more "hard science" discussion.

That is why I have both and always commend others to buy and read both.

cheers

 

[tuga]

I don't think that you've read Toole's replies or considered the concept of testing a single variable at a time carefully enough. Otherwise, one might look at the information presented already and conclude that nothing really matters in stereo, anyway, since everything rates relatively similarly. Similarly, one might actually believe that two pairs of speakers placed at different locations (is such a thing as "optimal position" even possible to define? What about toe-in?) in the room and equalized (how? Direct on-axis or at the listening position, what kinds of filters, effects on phase or delay, etc) can be compared in a valid way scientifically, as if multiple variables were not being conflated therein.

My question are:

Why should the spatial quality of a single speaker matter unless you only listen to a single speaker?
(What is spatial quality if not the interaction of the speaker's dispersion/directivity characteristics and the room boundaries? Can we perceive spatial quality in anechoic conditions?)

and

How can you conflate that a single speaker which rates highly in terms of spatial quality when listenened to in mono is objectively better than one which rates poorly when they rate similarly if listened (as intended) in stereo?


Optimal positioning is one which will produce the flattest response at the listening spot (and the listening spot may not be the same for every speaker design). There are two optimal positions for each speaker design: one for the long wall and one for the short wall setup.
Also different topologies require distinct positioning in the room: Allisons should sit against the front wall, the D&D should be place close to it, dipoles away from the front wall, omnis away from horizontal boundaries, corner horns in the corners, etc.


Now toe-in.
Let's look at the Dali example, which is hardly an off-beat, small-distribution manufacturer. The frequency response is flat at 30º by design and that is the listening axis as per manufacturer indication. But looking at the photos and vidoes of the mno shuffler we see that all speakers are evaluated at the listening spot without toe-in. There are people sitting off-axis which may make sense from a practicality perspective (you increase the sample pool a lot faster that way and more cheaply too) but having several people in the same room is not ideal, nor is listening off the ideal axis unless you are trying to determine how balanced the speaker sounds off-axis (which is not the same as evaluating spatial quality), in which case such assessment should not be conflated with the on-axis assessment.
Anyway, this is the 0º and 30º frequency response of the Dali Opticon 8:

A couple of photos from the shuffler(s):

 

[tuga]

The merits and demerits of mono evaluation are off topic and better discussed here:

Speaker Testing: why mono is better​

Speaker Testing: why mono is better

This topic of "why Amir tests speakers in mono" keeps coming up. I must have explained that a hundred times in text. Most of you probably know why. But I thought I do a video that covers the research and explains it all. Here it is: References...

www.audiosciencereview.com

 

[tuga]

I agree. The fact that stereo listening lifted the spatial scores of speakers that performed worse on that metric in mono, is not a logical reason to say stereo is better for judging spatial ability.

Let's look at this figure:

(If we ignore possible bias generated by each speaker's inherent frequency response and room positioning) perhaps you will agree that wide-directivity is responsible for the higher spatial quality rating.

Now in stereo conditions the Quad's rating improves significantly.
And the Quad/dipole speaker has one particular quality which is it's ability to reduce the level of early reflections from walls, floor and ceiling whilst simultaneously increasing the level of the late difuse sound coming from the front wall creating a sort of Reflection-Free Zone with reduced interference and increase sense of "spaciousness" and imaging sharpness:

 

[Newman]

Hi Tuga, IMO there is not enough data to jump to that particular, possible, conclusion. More of the data actually supports the conclusion that stereo simply dulls the ability to discriminate, compared to mono. Look at the data spread in stereo, of all 3 speakers. Same spread. If there were error bars they would overlap so much that the only possible conclusion on the spatial ability of the 3 speakers would be ‘same-same, now give me better data’. We can’t actually say the Quad improved its sense of space at all, while the data is suggesting that the test is not an appropriate test for that attribute.

 

[abdo123]

Let's look at this figure:

(If we ignore possible bias generated by each speaker's inherent frequency response and room positioning) perhaps you will agree that wide-directivity is responsible for the higher spatial quality rating.

Now in stereo conditions the Quad's rating improves significantly.
And the Quad/dipole speaker has one particular quality which is it's ability to reduce the level of early reflections from walls, floor and ceiling whilst simultaneously increasing the level of the late difuse sound coming from the front wall creating a sort of Reflection-Free Zone with reduced interference and increase sense of "spaciousness" and imaging sharpness:

Forgive me for my ignorance but wouldn't these dipole qualities also exist in the mono test as well?

Also how does a dipole deal with the ceiling reflections exactly?

 

[tuga]

Hi Tuga, IMO there is not enough data to jump to that particular, possible, conclusion. More of the data actually supports the conclusion that stereo simply dulls the ability to discriminate, compared to mono. Look at the data spread in stereo, of all 3 speakers. Same spread. If there were error bars they would overlap so much that the only possible conclusion on the spatial ability of the 3 speakers would be ‘same-same, now give me better data’. We can’t actually say the Quad improved its sense of space at all, while the data is suggesting that the test is not an appropriate test for that attribute.

What is the point of determining how good a speaker’s spatial performance is in mono if it will never be used as such?

And although I agree with the possible interpretation that stereo dulls the ability to discriminate spatial differences in my view such differences are only valid for single speaker listening and ignore the fact that a single speaker will be positioned very differently in the room from when used in a stereo pair. And because it is located further from the side walls in mono and close to them in stereo the spatial quality in mono cannot even be transferred to stereo.

My interpretation of the data is that the wide directivity speaker rates higher in terms of spatial quality in mono because when positioned equidistant to the side walls it will interact more with them than narrow or narrowing directivity ones.

 

[tuga]

Forgive me for my ignorance but wouldn't these dipole qualities also exist in the mono test as well?

The speaker is positioned in the middle of the room, far from the side walls, and the axis is perpendicular to the front wall.