The “Real” Speaker Preference Test

[levimax]

Woofer integrated close to the mid+tweeter enables higher XO frequencies such as 300-400 Hz. That is not a problem for FIR engine.

Continued off topic. I agree close co-locating subs with mains solves many problems. My question for you is the "standard" rule is to keep drivers within 1/4 wavelength of the crossover frequency. I have read some of your writings and you seem to recommend keeping drivers within 1 1/4 wavelengths. Can you clarify? Thank you.

 

[kimmosto]

you seem to recommend keeping drivers within 1 1/4 wavelengths. Can you clarify?

1/4 wave length or shorter is usually impossible between M and T. Next longer and possible c-c distance range giving smooth power averages i.e. SP and ER in vertical plane is 1-1.4 wave length at XO. Including ER is more valid to mid-field where vertical ER has significance. SP rules to much longer distances. Main idea behind all this is to avoid c-c = 1/2 wave length which causes the highest hump to directivity index i.e. the deepest dip to power response compared to on-axis. Vertical lobing to smaller vertical angles is the price you have to pay for smoother DI and power. That can be partly controlled with steeper XO or aiming main axis a bit up by delaying tweeter (assuming tweeter is upper). Up is more common direction for lobing problems assuming that listener does not sit in a spot all the time.
Suitable c-c is even more important if the speaker has other reasons for power dip at XO. Such as too large mid-woofer or/and too large and deep wave guide for the tweeter. Less problems if directivities of M and T compensate each others.

 

[levimax]

1/4 wave length or shorter is usually impossible between M and T. Next longer and possible c-c distance range giving smooth power averages i.e. SP and ER in vertical plane is 1-1.4 wave length at XO. Including ER is more valid to mid-field where vertical ER has significance. SP rules to much longer distances. Main idea behind all this is to avoid c-c = 1/2 wave length which causes the highest hump to directivity index i.e. the deepest dip to power response compared to on-axis. Vertical lobing to smaller vertical angles is the price you have to pay for smoother DI and power. That can be partly controlled with steeper XO or aiming main axis a bit up by delaying tweeter (assuming tweeter is upper). Up is more common direction for lobing problems assuming that listener does not sit in a spot all the time.
Suitable c-c is even more important if the speaker has other reasons for power dip at XO. Such as too large mid-woofer or/and too large and deep wave guide for the tweeter. Less problems if directivities of M and T compensate each others.

Thank you, does same apply to woofer to subwoofer?

 

[Arindal]

anytime an attempt is made to directly link product sales performance to quality raises a multitude of flags for me.

Certainly true, but I did not really see this particular claim being made. The initial question was rather about a link between sales figures of loudspeakers and their preference ratings in particular controlled tests. Note: preference ratings, not sound quality. I would expect a similar correlation as observed with mass-market portable speakers, for example.

Showroom sound" is a thing for a reason.

I am pretty skeptical when it comes to ´showroom sound´ in the sense of boosted bass and treble, really driving sales of expensive products today. Reason being, many contemporary pop music productions are already containing such ´showroom sound´ EQ curve, lots of effects like aural exciter, distortion, autotune distortion. Applying an additional treble boost EQ most likely will lead to annoyingly bright and harsh sound in an increasing number of cases. Same with lots of rock, hiphop and metal productions being subject to excessive compression.

Would assume that quite a number of potential buyers would prefer speakers that don´t sound annoying with their particular mix of genres. A bit like audiophiles shying away from any harshness.

 

[kimmosto]

Thank you, does same apply to woofer to subwoofer?

No. M+T only because that's commonly the only pair possible to locate close to c-c=1/2 wave length at XO which is the worst case scenario for smoothness of DI and SP.

Should also mention that c-c=1/2 wave length at XO could also be good if you really need higher DI and lower SP at XO range. That is possible for example with dipole and hyper-cardioid with electro-dynamic drivers because baffle (wider to cone) and other acoustical delays could create wider beam width close to upper XO frequency. As usually, almost everything is case/concept-dependent.

 

[RickS]

There was a already a thread as well from @Kvalsvoll with interesting information in his blog.

Thanks Tim!

That thread is still open and may be a more appropriate place for some of the off-topic posts in this thread.

 

[Floyd Toole]

I sense some confusion, misunderstanding and possible bias in some of this discussion. This happens because most people haven’t read the science story and get influenced by internet chatter and maybe by what they just heard. As most of you probably know I have just published the 4th edition of my book “Sound Reproduction” with co-authors Sean Olive and Todd Welti, and I just learned that like earlier editions it will be translated into Chinese. It is a long, detailed description of about 60 years of research – not a relaxing read compared to audio forums. I started the research in Canada, which ended while I was at Harman International, so common chatter implies that it was all done under the auspices of a for-profit company to benefit its own products and is therefore suspect.

A little “insider” perspective might help this discussion. It is a long read so it is attached.

 

[thewas]

A little “insider” perspective might help this discussion.

 

[Buckchester]

This ´carved in stone´ is a very good point. What is oftentimes referred to as ´scientific´ on this board, setting extremely high thresholds how controlled, blind listening tests should look like, is in my understanding a measure to prevent any other entity than Harman, to ever produce or publish any test result that would be accepted as ´scientific´, due to high costs of blind test facilities and staff to run it. While on the other hand disparaging all other tests conducted ever since as ´non-scientific´, ´biased´.

Interestingly, Harman apparently lost interest in conducting controlled tests on loudspeaker sound quality, according to Dr. Olive somewhen around 2012 (if I recall it correctly). Add these two things up, and you basically have some decades-old ´scientific findings´, based on existing products and measurement techniques of the days, which are frozen in time ever since, defended by everlasting exegesis, which must not and cannot be contested, will never be put to the test including practical findings and product solutions which evolved after the last wave of ´scientifically righteous´ products developed in the late 2000s. I personally very much would like to see some cardioids and line sources being put to a blind comparison test, not to speak of different speaker properties (you mentioned GD) in isolated testing.

The interesting question, having to do with the title of this thread, is: why? And where are these dominant products evolving from ´the science´, that was stopped some 15 years ago?



I agree, although I am pretty cautious when it comes down to predicting an outcome or clear audibility thresholds.

Interestingly, K+H (today: Neumann) conducted such tests, and invited recording engineers to take part, when launching their first FIR-controlled speaker in the early 2000s. If I recall it correctly, that product was named O500C, and included instantaneous implementation of GD for listening tests. The ´fullrange linear-phase mode´ was particularly interesting.



Can confirm that, they had the ability to to do that. For loudspeaker evaluation, particularly for judging tonality and comparing overall sound quality to competitors, according to Dr. Toole, mono testing was dominant, for several reasons such as better discrimination.

 

[illusonic]

Harman speaker research did not stop 2007:

https://www.lexicon.com/on/demandware.static/-/Sites-masterCatalog_Harman/default/dw1c32ffd5/pdfs/Lexicon_SL-1_System_Brochure.pdf

And some smart speakers (e.g. Apple Homepod) and soundbars (e.g. Bose) do similar things on a smaller scale. For example a single frontal speaker system for +/-90° envelopment. But home stereo doen't seem to attract enough consumers anymore. Most are happy with BT speakers/TV/soundbar. Best sound system is often in the car.

 

[Floyd Toole]

Harman speaker research did not stop 2007:

https://www.lexicon.com/on/demandware.static/-/Sites-masterCatalog_Harman/default/dw1c32ffd5/pdfs/Lexicon_SL-1_System_Brochure.pdf

And some smart speakers (e.g. Apple Homepod) and soundbars (e.g. Bose) do similar things on a smaller scale. For example a single frontal speaker system for +/-90° envelopment. But home stereo doen't seem to attract enough consumers anymore. Most are happy with BT speakers/TV/soundbar. Best sound system is often in the car.

This is not a product of research, it was product development - a big difference. It is an exercise in array design. I heard prototypes while I was at Harman, so it has been around for a while. I was part of the effort to get the designer into the US from Switzerland - a brilliant engineer. It was an excellent design, extremely complex and expensive so there were long debates about whether it would get to market. If it has, at $38k USD/pr it is not a mass-market product.

Here is a spinorama of one of the forward-firing prototype arrays from 2005 - it is an excellent loudspeaker, so arrays work - notice the directivity control down to about 300 Hz. Well done - but not cheap.

Arrays have existed for decades, but small inexpensive DSP and class D amps have unleashed new capabilities. They are definitely used in sound bars and some smart speakers to steer sounds towards reflecting surfaces for more expansive space illusions.

 

[Duke]

A little “insider” perspective might help this discussion. It is a long read so it is attached.

I appreciate your taking the time to write this, and especially the attachment.

Quoting from your attachment:

"I soon suggested that a large company like Harman (then about $500M sales) could afford
to subsidize a small research group to continue the research that had been so successful in
Canada. Amazingly, my colleagues agreed, and more remarkably agreed that the results of
the research could be discussed and published openly in the scientific community, the AES.
The results of the research would be knowledge disseminated to the world, not products,
although the knowledge could logically be used by design engineers associated with the
brands...

"As for me, I am eternally grateful to have been well paid to pursue my interests for my
professional life – nobody ever told me what to do, it was research. The Canadian
government and Harman International have my eternal gratitude for injecting large sums of
money into a scientific examination of the acoustics and psychoacoustics of loudspeakers
and rooms. It was productive, it is all in the public record – no company secrets – and it
won’t go away."

As one of the countless many who have benefited enormously from the selflessness of you and Harman, a big THANK YOU!

 

[RexrothPigeon]

The notion that because Toole, et al.'s studies happened some time ago and nobody has done studies to demonstrate that they're wrong we should then throw out those studies and assume something contradictory to them based on (tortured logic? undocumented tests for which no data are available because reasons? the unsubstantiated testimony of someone on the internet?) is deeply contrary to the scientific spirit of this forum.

If someone produces (verifiable, replicable) science that undermines Toole, et al.'s findings or that reveals flaws in Klippel NFS measurements, then we will have a new understanding, but that hasn't happened and I don't see any of the anti-Toole peanut gallery doing the work necessary to potentiate such an outcome.

 

[illusonic]

This is not a product of research, it was product development - a big difference. It is an exercise in array design. I heard prototypes while I was at Harman, so it has been around for a while. I was part of the effort to get the designer into the US from Switzerland - a brilliant engineer. It was an excellent design, extremely complex and expensive so there were long debates about whether it would get to market. If it has, at $38k USD/pr it is not a mass-market product.

Here is a spinorama of one of the forward-firing prototype arrays from 2005 - it is an excellent loudspeaker, so arrays work - notice the directivity control down to about 300 Hz. Well done - but not cheap.

Arrays have existed for decades, but small inexpensive DSP and class D amps have unleashed new capabilities. They are definitely used in sound bars and some smart speakers to steer sounds towards reflecting surfaces for more expansive space illusions.

Not sure where to draw the line between research and development. The specialty of the SL1 is the ability to steer the beam direction in the horizontal plane and to adapt beam width up to 360° radiation. All controllable for 2 speakers through an app. So this is not only an array but an acoustically highly sophisticated one that allows such flexibility just by adapting beamforming filters. To me it also means research to come up with an idea for a truly new product and to work on it through various prototypes until it becomes clear that what one envisioned is technically feasible. The development part would then be to finally optimize implementation such that it becomes a product. If the former is not research, the question arises how much research is carried out in R&D departments around the world. Another question would be if one can be a serial inventor holding numerous valuable patents without being a researcher. In many cases actual new knowledge is comprised in patents. It seems to me that there is the pure knowledge generation part of research and the pure product execution part of development but no clear line between them.

 

[delta76]

Does anyone have any actual facts and figures to show if speakers that have been designed to apply the Harman research findings have actually made any significant sales inroads with the buying public? The ultimate Preference Test is whether people are parting with their hard earned money when given the option and ability to compare in the marketplace. Or are the top selling names in the marketplace (whoever they may have been) still the same?

the comparison is only meaningful for two pairs that are made by same manufacturer, look fairly similar, priced similar, one is Harman curve optimized and one is not. Of course no manufacturer is going to do that.
We at ASR are strongly influenced by the measurements, but it's not the only/deciding factor. Brand, look, price, availability can all play a part.
For the wider world, they are the deciding factor not the measurements.

But manufacturers like Ascilab is selling good chunks of speakers, so that says there is a market for that!

 

[Floyd Toole]

Not sure where to draw the line between research and development

Yes indeed it is not always a clear line. One does not need a white lab coat and a PhD to do research - I have known some technicians with more savvy than the engineers they worked for, and some engineers who by virtue of their development work deserved PhDs. But if the target is defined, and there are trusted measurements that describe progress towards the target, I consider it engineering development.

In this case there was no question to be answered about the goal: uniform directivity, variable from omnidirctional to something resembling a forward firing loudspeaker, in some number of steps. Of course the long established requirements of no resonances and uniform directivity at frequencies above transition would apply. From that point on it is an engineering exercise - a challenging one - with measurements as the proof of success. Any listening that was done easily validated the result. From there on the user interface, the moveable listening position, etc. production engineering, had to be done. This came near the end of my full-time employment as VP but I experienced demonstrations of all those features. They worked. I sometimes wondered why anyone would spend so much money to avoid sitting in the symmetrical stereo seat - but I digress: that is a marketing feature, a differentiation.

 

[Mnyb]

Ok we should stop calling it "harman research" as the work was done under several managements . I understand that there is need for a shorthand .

I understand that there are objections . And the research may need to continue and might get even more refined results and move . You might not even consider it complete.

That is not an argument for going back to the wild west of speakers designing and just wing it due to some personal pet theory .

There are constant pot shots by some posters at the research , instead present clear arguments supported by data . Don't only say "this is wrong" also tell us what's right please

 

[RDoc]

Could you list all variables and their targets included in 'Harman principles'?

Look at the speaker preference ratings and how they are calculated. For a deeper dive read Floyd Toole's book(s)

 

[Sokel]

Look at the speaker preference ratings and how they are calculated. For a deeper dive read Floyd Toole's book(s)

I believe that his VituixCAD shadow PIR, LW and SP reflects it.
At least the Sean Olive's one.

 

[Floyd Toole]

I believe that his VituixCAD shadow PIR, LW and SP reflects it.
At least the Sean Olive's one.

I believe that his VituixCAD shadow PIR, LW and SP reflects it.
At least the Sean Olive's one.

True, but it is not the individual shapes of these curves that convey the key fact, it is the observation that evidence of resonances - the most audible problem in loudspeakers - is mostly revealed in large spatial averages, which all these curves are. If a spike, bump or hump persists from on-axis through listening window and early reflections (closely equivalent to PIR that is dominated by it), to sound power (the ultimate spatial average) it is definitely a resonance. Smoothness and flatness in the direct sound is important (on-axis and listening window), and smoothness in the other larger spatial averages, all of which are indications that the loudspeaker lacks resonances. The actual shapes of the off-axis curves is of value, but definitely secondary, as they relate to later arriving reflected sounds.

So, have a look at Sean's findings and the words smoothness (SM), flatness (AAD) and narrow-band deviation (NBD) appear in front of the important curves. The model is finding resonances, but doesn't use the word.

Here is an example of a loudspeaker that would do badly in the model with the resonances identified. Highly rated loudspeakers exhibit smooth curves, beginning with a flat direct sound.