Interview about Measurements with Pearl Sibelius Speaker Designer

[ctrl]

That is fine in theory, but very few audiophiles use EQ. and many audiophiles who listen to vinyl refuse to pass their precious analogue signal through an A/D and D/A stage

Cursed be all audiophiles! In this case, a well-designed speaker with subjectively good sound is likely to be a good choice.

 

[Vacceo]

It is a fallacy that measurements need to provide the full picture to be useful or even critical. Nothing in life is produced that way. Yet, technology advances and adoption is massive in scale. Measurements easily quantify bad products. And put the wind behind performance of a speaker when they are good.

Ultimately audio has a broken architecture in that the capture process does not comply with any standard. So there will never be the case that you fully know what is right when it comes to tonality. Fortunately immense enjoyment can be had without achieving that perfection.

I have been in Harman speaker shuffler room twice. It is a great experience. Fortunately you don't need to be there to follow the research that connected the preferred speaker there with measurements. So many people are blindly buying speakers, headphones and IEMs as a result of our objective work and vast majority are quite happy. This is all we can hope for.

It is very possible that I will never experienced the shuffler room. It is also quite possible that I will never experienced space travel either.

But luckily I can know about both. You do the heavy lifting (sometimes, as with the Genelecs, quite literally) so I can have very good data to make an informed decision. And that is exactly what science does: provide you with good enough data to make choices.

Plus, I know I'll never be able to afford you a pair of tickets to adjust my listening room, and you don't even try to sell me this service, so the least I can do is be grateful for your answers, and send cheers every once in a while to you and all those who help me to learn.

 

[tuga]

Wasn't the birth of the "High End" when tone controls left the preamp.

What's a preamp?

 

[Mr. Widget]

What's a preamp?

Oh, you use a passive volume control?

 

[tuga]

Oh, you use a passive volume control?

A single box amplifier.

 

[Mr. Widget]

A single box amplifier.

To be clear I was joking... I was going to the next level of loss of control and audiophiledom.

Seriously though, I assume a single box amp is another term for an integrated amp. Is that phrase generational or regional?

 

[tuga]

To be clear I was joking... I was going to the next level of loss of control and audiophiledom.

Seriously though, I assume a single box amp is another term for an integrated amp. Is that phrase generational or regional?

I was pulling your leg.

 

[Mr. Widget]

We are going to have sore legs if this goes on much further.

Back to your post #38, having seen so many "smily faced" graphic equalizers back in the day... I wonder what the ratio of positive outcomes to failures is with modern PEQ controls? Ctrl's suggestion of just going with a well designed speaker with subjectively good sound is the right answer for most of us.

 

[Steven Holt]

keep an open mind and listen as well as measure.

I have read many of Amir's reviews and measurements, and although I understand roughly 50 pct. (still learning), I can tell u that Amir listens closely to any speaker, and amp, that he reviews. He tells u so at the end of the review, 'subjective listening test'. As for an open mind, he's an engineer by training. They never make up their mind ahead of time (and he has been surprised by some gear). The proof is in the pudding -- or it isn't.

 

[Hans Lange]

It is quite possible that a speaker with very good measurement results sounds subjectively worse to you than a less well-designed speaker with better fine tuning.
The well designed speaker should be relatively easy to fix with a little EQ - the opposite case would be much more difficult or impossible to realize.
Therefore, it is always an advantage if the speaker designer has so much experience and knowledge to realize both (good subjective sound and good objective measurements), because then the speaker is most flexible to use.

More details about the design process of a loudspeaker and why complete measurements are important (using the example of SP and PIR) can be found here in the discussion of an interview with the Dynaudio designer.

One problem, however, is how "better measurement" is defined. Unfortunately many equate a flatter on-axis frequency response with "better", which is total nonsense if the off-axis behavior of the speaker is not known (if we exclude obvious things like a 10dB hump or similar).
Also in the conversation, it seems like both of them are always talking about on-axis FR, which is not enough to describe a speaker, especially a single drive unit speaker.

One can point this out a hundred times (this was one reason for the introduction of the CTA-2034-A standard) in the next discussion it will be said again there is a flat measuring speaker that did not sound good! Gotcha!




From the designer's point of view, this is certainly an understandable position.
He also said:

I found this a bit strange, because after the selection of the driver, the real design work begins and you simulate which enclosure shape and driver position is as optimal as possible and to be able to predict how which changes will have an impact. Only then the first prototype is built. But maybe he just forgot to mention this.

and compared two completely different things:

How wrong this analogy is does not need to be explained further.

There is already a slight antipathy to measurements expressed, also in his anecdote about the design of a speaker with the best possible flat frequency response that sounded like rubbish.
This all sounds a bit like an old-school designer for whom nothing else exists besides the on-axis FR which serves as a reference point and one can only trust your own ears - which is not wrong, but does not tap into the potential of today's measurement, software and simulation possibilities.


To demonstrate this, we can simply simulate how the Sibelius will probably measure from about 300-400Hz.
To do this, take the manufacturer's on-axis half-space measurement of the Mark Audio Alpair M10 Gen. 3 (should correspond more or less to the driver of the Sibelius) and simulate the influence of the baffle under free-field conditions (using VCAD software).
View attachment 236593

The low frequency behavior is not simulated. The Sibelius seems to use a transmission line concept to extend the low frequency reproduction. That is why a comparison is made in the interview of organ pipes, which are also "lambda/4" resonators. Since unwanted resonances from the TL port are always transmitted as well, the FR will turn out more wavy in the low-mid frequency range - beside the extended low frequency range.

Simulated FR for Sibelius deg0-20-40
View attachment 236591
With this concept, the on-axis frequency response is not decisive for the sound impression in the listening room. If the LS would be tuned to a flat on-axis FR, the LS would sound lifeless and dull, because then SP and PIR would drop too steeply to high frequencies from about 4kHz. With complete measurements, we could assess how even the radiation of the loudspeaker is and locate possible frequency ranges to be observed in particular.

The simulations also help assess how changes in enclosure shape or driver position are likely to affect the system (without building a prototype for every design).
normal width or 300mm width
View attachment 236593 View attachment 236603

To my understanding this 4" driver has the virtue of a rising HF response. Being a one driver design in terms of precise soundstage it's a convincing product when listening on axis. In terms of polar response or directivity, 34dB loss at 10 kHz @-20° is a bit too much. Of course some loudspeakers may be tuned or voiced by listening to them...measurements however should be given a bit more attention, that's at least our policy. It's my first post here, interesting topics are discussed deeply and in detail...very intriguing;

 

[sarumbear]

If the speaker sounds "fantastic", then of course the dip should be left in place. The only important thing is by evaluating the crossover simulation tool (e.g. VCAD), to make sure that it does not show any indicators of a "colored sound".

First of all what is "fantastic" or "coloured" sound? They are subjective adjectives. Their meaning differ from person to person.

Secondly, if a speaker designer, whom I assume to have some sort of education relating to the subject, finds a speaker that he thinks sounds fantastic but doesn't measure well, he should then conduct a research and find the reason. This is not rocket science nor we are expecting to re-write scientific research. It is plain old engineering. If you cannot find why the sound of a speaker with a dip on its response sounds "fantastic" then there are only two reasons.

1- You are not a good designer/engineer
2- You are not an experienced listener

End.

 

[tuga]

First of all what is "fantastic" or "coloured" sound? They are subjective adjectives. Their meaning differ from person to person.

Secondly, if a speaker designer, whom I assume to have some sort of education relating to the subject, finds a speaker that he thinks sounds fantastic but doesn't measure well, he should then conduct a research and find the reason. This is not rocket science nor we are expecting to re-write scientific research. It is plain old engineering. If you cannot find why the sound of a speaker with a dip on its response sounds "fantastic" then there are only two reasons.

1- You are not a good designer/engineer
2- You are not an experienced listener

End.

I am not disagreeing with you in principle but this idea that any speaker manufacturer can "conduct a research and find the reason" is absurd. Harman, in spite of its wealth and Toole's praiseworthy efforts, tried and failed.

 

[sarumbear]

Harman, in spite of its wealth and Toole's praiseworthy efforts, tried and failed.

Where did they fail and how do you know about it?

 

[tuga]

Where did they fail and how do you know about it?

It's been discussed in a post you started about the preference rating.

 

[sarumbear]

It's been discussed in a post you started about the preference rating.

I appreciate if you refresh my memory.

 

[amirm]

I am not disagreeing with you in principle but this idea that any speaker manufacturer can "conduct a research and find the reason" is absurd.

Why? They can't be bothered to run a controlled test between two designs with a dozen listeners? Instead we should just take their contraraian word for it?

 

[witwald]

...this idea that any speaker manufacturer can "conduct a research and find the reason" is absurd.

Wouldn't it be an entirely reasonable expectation that any of today's loudspeaker manufacturers will take the necessary time to conduct a design exploration when bringing a new product to market? After all, each and every combination of drivers chosen for a particular design can end up using different crossover networks while still achieving a workable, marketable solution. Many loudspeaker reviews, as well as the attendant brochures supplied by manufacturers, allude to the "voicing" of loudspeakers. Is not the process of "choosing" the desired "voicing" a method of "conducting research"? If it is, then all manufacturers do it, to one extent or another.

 

[tuga]

Why? They can't be bothered to run a controlled test between two designs with a dozen listeners? Instead we should just take their contraraian word for it?

We should not take their word for it, no.

 

[tuga]

Wouldn't it be an entirely reasonable expectation that any of today's loudspeaker manufacturers will take the necessary time to conduct a design exploration when bringing a new product to market? After all, each and every combination of drivers chosen for a particular design can end up using different crossover networks while still achieving a workable, marketable solution. Many loudspeaker reviews, as well as the attendant brochures supplied by manufacturers, allude to the "voicing" of loudspeakers. Is not the process of "choosing" the desired "voicing" a method of "conducting research"? If it is, then all manufacturers do it, to one extent or another.

A loudspeaker can be successful because it looks good or the brand has strong appeal or the sound presentation (set of distortions) appeals to some people.
Speaker manufacturing is not a race to the best measuring trophy.
It’s a business, which obviously needs to make a profit, and the sound needs to please the potential owners as well as the designers.

 

[Mart68]

A loudspeaker can be successful because it looks good or the brand has strong appeal or the sound presentation (set of distortions) appeals to some people.
Speaker manufacturing is not a race to the best measuring trophy.
It’s a business, which obviously needs to make a profit, and the sound needs to please the potential owners as well as the designers.

I don't think it is that simple. Obviously the preference score is flawed and no-one claims it isn't.

But the real world clashes with the principle - people tend not to audition in their own acoustic, and even if they do, they audition with showcase recordings that will sound good on anything.

Once they live with the speakers for a while they discover that a lot of recordings sound bad on them. So they will complain about how poorly produced 1970s rock music is. They will then start chasing 'synergy' with new amp, new DAC, new cables, foo devices etc. The original purpose (enjoying music) gets forgotten.

Buying a speaker that measures well and sticking it in a good acoustic is still the best route to long-term satisfaction with the sound quality across all types of programme.