Andrew Jones’s new speaker brand - Jones and Cerreta

[Curvature]

The sound power shows that they will definitely sound different

Different, but one will not likely be preferred over the other given the preference scores:

• 420: 6.39
• Blade 2: 6.25

If within 0.7 the score difference is not considered significant.

More on topic, I think that demonstrates an important point about listening. Small, audible differences aren't necessarily significant and may not need to be part of design decisions. Without a consistent interpretive overlay the designer's time spent listening may be, in some if not most cases, a waste of time.

Speaker design these days, for small companies, is a combination of simulation, measurement and listening. The main reason is that the errors are gross at each stage. A typical designer will use basic simulation tools (but not full FEM), check the simulation with quick measurements and then listen at the end. Introducing expert simulation slows the process considerably and adds expense, as does automating measurement with an NFS (vs. a few basic curves). I can see how, in a situation with significant information gaps, including not being familiar with perceptual research, it is simply more "efficient" to listen.

The reason these companies are at all competitive is that the initial barriers to entry in terms of knowledge aren't very high and initial investment can be kept low, particularly when "outsourcing" part of the labour to driver manufacturers and measurement labs.

It only starts making sense to completely remove listening from the work when you have control over more or less every element of the design process, which sets the knowledge requirements and necessary bar for investment much higher. As I said earlier, the main design conundrum is interpreting small changes or deviations, particularly in the frequency response. More specifically, Jones has spoken a lot about trying to trade on axis vs off axis response by considering where a listener is expect to sit and position their speakers (flat against the wall vs. not, firing directly out vs. toed in). This complicates the situation by trying to anticipate listener actions, including whether or not they will EQ to compensate for the room or dial in a tilt for taste.

I would push for a few things: more rigorous, controlled research and higher availability of measurements, and more effort to produce higher-quality measurements. I honestly don't care how a designer does their work apart from trying to understand the realities of the process. My expectations are more about the result and transparency.

 

[thewas]

Different, but one will not likely be preferred over the other given the preference scores:

• 420: 6.39
• Blade 2: 6.25

I didn't write anything about preference, the single number score like Toole says anyway is very limited and only sufficient to show differences between good and poor designs and people should rather look at all curves instead.

 

[a4eaudio]

So there is every reason (and nothing to do with 'my being so confident', as some are so quick to assert and others equally quick to 'like') to conclude that, even if Jones has been conducting controlled listening tests for the past 20 years that they would mislead him. Add to this the high likelihood that he is doing it sighted, and it amounts to nothing less than a mistake

I have no reason whatsoever to believe this. In fact, the only evidence I have refutes it. There are numerous speaker designers over 60 (and 70) that are designing great speakers. We have strong evidence that AJs listening tests have not misled him - this is not a theoretical debate, he has actually designed speaker that have been listened to and measured. I wish I could be misled like him.

All of the arguments about "sighted" voicing seen completely irrelevant to voicing a speaker. Voicing a speaker would typically involve fast A/B switching between a change in a component or two, or maybe different acoustic slopes, etc. The designer knows what his speaker looks like. What does "sighted" even refer to when sitting in a room A/B'ing a change in a resistor that you can't even see? (I wouldn't be surprised if most designers close their eyes when A/B'ing to help concentrate, so they are probably not "sighted" anyways, even though it seems meaningless.)

 

[SigbergAudio]

I think we are undervaluing the context we are talking about. We know that we are prone to bias on sighted listening. A speaker designer listening to his own speaker for hundreds or thousands of hours while developing it, is however a pretty different situation than a guy walking into a showroom and somehow preferring Zu Audio over Kef.

There can in theory of course be bias involved during development as well, but given the length and amount of listening combined with the fact that many of the choices don't really have any downsides from a manufacturing perspective (no need to have a bias), it's not unreasonable to assume that the developer is actually able to evaluate the difference.

As discussed in earlier threads, the fact that ears are not 100% reliable and we are prone to bias, does not mean we cannot distinguish differences in any circumstance. Taken to the extreme, we should all have 10USD bluetooth radios for our listening purposes.

 

[Newman]

It's interesting how much you know about Andrew Jones.

I'm going with the odds in the absence of evidence. That's the rational approach.

Very happy to be shown evidence to the contrary. Got any?

Yes, it's a natural pattern to see since I disagree. In the same way there are clear patterns in what you agree and disagree on.

So, which part of Toole are you disagreeing with, and why? What evidence supports your disagreement? Because, Toole, who I am agreeing with, bases his views on valid evidence.

You are trying to reframe this as agreeing and disagreeing. I'm not fooled.

 

[SigbergAudio]

I'm going with the odds in the absence of evidence. That's the rational approach.

Very happy to be shown evidence to the contrary. Got any?

I haven't seen any evidence that he has anything but a very rational and measurement led approach to speaker designing, so I am not sure what to show you.

So, which part of Toole are you disagreeing with, and why? What evidence supports your disagreement? Because, Toole, who I am agreeing with, bases his views on valid evidence.

I don't necessarily disagree with Toole, to my knowledge. Does he suggest that speaker designers shouldn't use listening as a means to evaluate their speakers during development (in combination with measurements, obviously)? I haven't seen him mention this.

You are trying to reframe this as agreeing and disagreeing. I'm not fooled.

So what is it if it's not agreeing and disagreeing? If I said the moon was made of cheese, I'm sure you'd disagree? We may have different definitions of the word.

 

[Curvature]

I didn't write anything about preference, the single number score like Toole says anyway is very limited and only sufficient to show differences between good and poor designs and people should rather look at all curves instead.

I added the point on preference to address the question of how to interpret small differences, whether or audible or measurable. The preference score is an interesting tool for doing so if you follow the research in detail. Not perfect by any means, but good for analytically representing the differences in measured data.

 

[Newman]

As Toole has explained, he chose to opt out of participating in experiments as a listening subject around the age of 60 because, unlike you, he was aware of the much deeper and more profound impact, of his (normal, age-related) hearing loss, on his ability to assess loudspeaker sound quality in numerous ways that matter.

It's absolutely not about 10-20 kHz.

Of course, unlike Jones, Toole's deep experience in listening to flaws in loudspeakers is supplemented by a deep knowledge of human psychoacoustics. Plus, he correlated his knowledge of the specifics of age related decline with his own measured loss in ability to discern loudspeaker flaws in controlled experiments.

So there is every reason (and nothing to do with 'my being so confident', as some are so quick to assert and others equally quick to 'like') to conclude that, even if Jones has been conducting controlled listening tests for the past 20 years that they would mislead him. Add to this the high likelihood that he is doing it sighted, and it amounts to nothing less than a mistake.

I have no reason whatsoever to believe this [last paragraph]. In fact, the only evidence I have refutes it. There are numerous speaker designers over 60 (and 70) that are designing great speakers. We have strong evidence that AJs listening tests have not misled him - this is not a theoretical debate, he has actually designed speaker that have been listened to and measured. I wish I could be misled like him.

(my bold above)
That 'evidence' is on the weakest evidentiary tier. 'Old people designing speakers that measure well is evidence of their sighted listening skills'???? That is an indefensible argument if valid experimental evidence is the standard. If they are lucky, their wandering around down the rabbit hole with sighted listening (after achieving excellent measurements) won't do too much harm. Fancy capacitor of the same value. Silver wiring. Different damping material. Many other possibilities.

All of the arguments about "sighted" voicing seen completely irrelevant to voicing a speaker. Voicing a speaker would typically involve fast A/B switching between a change in a component or two, or maybe different acoustic slopes, etc.

Fast A/B switching in sighted listening is still subject to the sighted listening effect, it doesn't change any of the failings of sighted listening.

The designer knows what his speaker looks like.

He or she also knows what change he or she hoped to achieve with the tweak that has been made, and what sonic difference it 'should' create, or is hoped to create. Bias.

What does "sighted" even refer to when sitting in a room A/B'ing a change in a resistor that you can't even see? (I wouldn't be surprised if most designers close their eyes when A/B'ing to help concentrate, so they are probably not "sighted" anyways, even though it seems meaningless.)

Err, seriously? 'Sighted' means knowing, not having eyes open. It is synonymous with "lacking controlled experimental conditions".

cheers

 

[kharan]

I think we are undervaluing the context we are talking about. We know that we are prone to bias on sighted listening. A speaker designer listening to his own speaker for hundreds or thousands of hours while developing it, is however a pretty different situation than a guy walking into a showroom and somehow preferring Zu Audio over Kef.

There can in theory of course be bias involved during development as well, but given the length and amount of listening combined with the fact that many of the choices don't really have any downsides from a manufacturing perspective (no need to have a bias), it's not unreasonable to assume that the developer is actually able to evaluate the difference.

As discussed in earlier threads, the fact that ears are not 100% reliable and we are prone to bias, does not mean we cannot distinguish differences in any circumstance. Taken to the extreme, we should all have 10USD bluetooth radios for our listening purposes.

I'm going to push back against this. It's often said in the audiophile world that the biggest source of bias is money and that having bought the speaker and then evaluating it makes it the most biased evaluation.

But an even bigger source of bias is familiarity, emotional involvement, and knowledge of what it's supposed to do. I've heard so many variants of this over time, a genius designer who has found a new way to improve upon the standard bass reflex calculations and measurements with some novel transmission line or bass port design. He doesn't measure anything, or the measurement isn't done with the necessary precision. Yet, he knows it sounds better, and when he listens to it, it obviously sounds better to him. While there is an actual sound difference, it's ego and bias that make him evaluate that difference as an improvement rather than just a variation.

 

[Frank Dernie]

Neumann simulate their drivers and speakers and then measure them when the prototypes are finished. Listening is not part of the decision process. The main thing is rigorous simulation, testing and measurement. Their business and resources are of course very different since they mass produce a highly competitive and tight product list.

Jones works as a consultant or something like it. He designs drivers and the rest, and the way he makes compromises between significant defects is through the listening assessment. I would imagine at TAD the development and manufacturing allows more budget per speaker to introduce driver and other changes to improve the performance than elsewhere, so the results are better. But those are extremely expensive speakers relative to ELACs. The economics of these companies are strange in that sense. I think the only reason they survive is because technical assessment is not part of the mindset of the average customer. Reputation rules and allows the market to be inefficient. How good are the drivers Jones makes relative to offerings by Tymphany or Scanspeak? I would imagine they are competitive but need to be fairly cheap to make to justify customization, which means compromises need to be made.

Neumann don't allow their technical targets to sway so much. The on axis flatness must be prioritized and off axis must be consistent with every other speaker in the line. The speakers must be linear with a very large range of signals. And each one off the line has to match as tightly as possible to the one before. Again, they have the resources to do this kind of work and are building on the knowledge of Klein & Hummel before them.

Erin said in that interview he's never seen a speaker with an on-axis response within 0.5dB deviation. The KH150 is within 1dB. https://www.spinorama.org/speakers/Neumann KH 150/ASR/index_asr.html And then the pair deviation is within, well, whatever this is: https://www.audiosciencereview.com/forum/index.php?threads/neumann-kh150-review.64775/

View attachment 555985

Recently I realized that the KEF Blade 2 Meta (released 2022, USD 14k) has nearly the same -6dB contour as the KH420 (released 2014, USD 5k):

KEF
View attachment 555992

Neumann
View attachment 555988

The economics allowing this level of disparity in price vs. performance are just incredible.

The J&C Troubadour speaker (USD 34k) looks like it uses the same basic coaxial driver design as the Mofi speakers, changing out the motor. I'll bet that the overall frequency response and directivity are largely similar, though tighter below 1kHz, and that they play a little louder with lower distortion compared to the 888 (USD 5k) or the V10 (USD 8k).

Edit: Error in price as noted by @thewas.

The more recent Genelec, and probably Neumann, active speakers with built in DSP have individually applied frequency response correction at the final inspection stage.
This is very sensible and probably a lot cheaper and easier than any method hitherto.

 

[Curvature]

The more recent Genelec, and probably Neumann, active speakers with built in DSP have individually applied frequency response correction at the final inspection stage.
This is very sensible and probably a lot cheaper and easier than any method hitherto.

An approach I wish were standard.

 

[Frank Dernie]

How I interpret it is that he has seen other people's measurements of his speakers with Klippel NFS and the measurements 100% match his own. He would be spending 150K to get what he already has, it's not that he can't afford it in the absolute sense it's that he can't afford throwing 150K in the wind.

My interpretation was that with the huge warehouse he said he tested in and gated measurements he is one of the pioneers of he actually doesn't need a NFS to interpret the data, though I imagine an automated system would be much easier..

 

[SigbergAudio]

I'm going to push back against this. It's often said in the audiophile world that the biggest source of bias is money and that having bought the speaker and then evaluating it makes it the most biased evaluation.

But an even bigger source of bias is familiarity, emotional involvement, and knowledge of what it's supposed to do. I've heard so many variants of this over time, a genius designer who has found a new way to improve upon the standard bass reflex calculations and measurements with some novel transmission line or bass port design. He doesn't measure anything, or the measurement isn't done with the necessary precision. Yet, he knows it sounds better, and when he listens to it, it obviously sounds better to him. While there is an actual sound difference, it's ego and bias that make him evaluate that difference as an improvement rather than just a variation.

Fair point, if you change something and don't measure at all, or measure improperly, just verifying it by listening - that's not a great approach. But this is at least not how I work, and I suspect Andrew Jones doesn't either. Any change is both measured and listened to. Sometimes it's easy to correlate, other times not. Sometimes I'm unable to find a clear reason in the measurements (this can sometimes be difficult even though it exists, like excessive energy off-axis, or a cabinet resonance or whatever else). Then I might have other people listen without knowing if or what has changed.

When building a commercial speaker (rather than DIY) it also needs to be tested in a number of rooms to see how it behaves with different placements, room size, etc. Obviously no speaker can be perfect for all situations, so everything is to some extent a compromise. Luckily modern speakers + modern amps/DACs/processors allows the end user to compensate to some extent for the room.

Currently I have changed a crossover point from 600 to 835hz on one of our speakers. Looking at the directivity, it looks slightly better. So I guess I should be inclined to think it sounds better as well. So far I think it sounds very similar, and it may only be an exercise in making the measurements look ever so slightly better. Not sure if it will be implemented or not.

 

[a4eaudio]

He or she also knows what change he or she hoped to achieve with the tweak that has been made, and what sonic difference it 'should' create, or is hoped to create. Bias.

If a designer wanted to address sibilance, he would try two different values of resistors. He does this because he knows the sonic differences caused by resistors. He could blind A/B compare the two resistors. Note, there is no bias because he doesn't actually care/have a preference over a green 4 watt resistor and an identical green 5 watt resistor.

 

[a4eaudio]

...When building a commercial speaker (rather than DIY) it also needs to be tested in a number of rooms to see how it behaves with different placements, room size, etc.

A large number of DIYers do this too, you just need to be fortunate enough to have access to a variety of rooms to work in.

At a recent DIY event there was an active speaker that was switched between an LR2 xo and LR4. They sounded different. There is no scientific/objective "better" speaker in this case. The only way to make a decision is to listen with human ears.

 

[SigbergAudio]

A large number of DIYers do this too, you just need to be fortunate enough to have access to a variety of rooms to work in.

Sure, I was thinking in a situation where one was making a speaker only for him/herself, there's not necessarily any need to test it elsewhere, you can even treat the room and speaker as one to some extent if you want, as the current room-speaker-system is the only place the speaker has to perform. Unless you plan to move of course.

 

[IPunchCholla]

If within 0.7 the score difference is not considered significant.

Aren’t the error bars +- 0.7 for each speaker? Meaning if they are within 1.4 they are not predictive of preference?

 

[Curvature]

it also needs to be tested in a number of rooms to see how it behaves with different placements, room size

I don't see any need for this at all.

At the end of the presentation Jones talked about judging a competition where the test songs are sent to the contestants in advance, and the contestants inevitably tailored their speakers for those songs for the win, since judging was by listening.

Speakers are general-purpose devices at the end of the day. I don't think it makes sense to compromise that by manipulating the FR in some way, on or off axis, based on a limited set of experiences or considerations, even assuming that the designer is good at controlling the circumstances of listening and coming to consistent judgments.

 

[Curvature]

Aren’t the error bars +- 0.7 for each speaker? Meaning if they are within 1.4 they are not predictive of preference?

Sure, if taking one speaker. The 0.7 is relevant when comparing two different speakers.

The way I interpret the score:

1. Vanilla: All factors important.
2. With sub: Bass differences removed.
3. With EQ: On-axis/listening window differences minimized, bass differences intact.
4. With EQ, with sub: Combination of 2 and 3. I.e., only the directivity is assessed.

 

[egellings]

What is the advantage of using a field coil over using a permanent magnet in a speaker driver?