Andrew Jones’s new speaker brand - Jones and Cerreta

[Sigberg Audio]

A few points:

1. We are at point where most people have to buy a speaker sight unseen. As a manufacturer, you better be sure what you design has high chance of acceptance by these customers. The designer tuning it by ear and with some music in some room is not going to get you there.

This is a slight strawman, while I'm sure there are designers out there that tune exclusively by ear, that's not what we're discussing here.

3. Proper listening test requires four or ideally 5-way blind, level matched test. This way, it is easier for listener to detect what speaker is out of place in tonality. Simply adjusting one speaker sighted, is not going to result in anything useful.

While I know the story being told on ASR is that ears are completely unreliable, this simply isn't true. Sure there are lots of bias involved, but that's not the same as it being useless. Harman themselves talk about trained and untrained listeners, and have developed software to help you become better at discerning changes in the sound. Does this ability go completely out the window unless you're in a double blind situation? No, it does not.

4. A much easier substitute for #3 is to use EQ to try out different approaches to response. That is what I did in this test:

I created a filter for that peak at 4 kHz, and AB tested myself and my son. This kind of testing can easily be blind and repeated. It was through this testing that I gave feedback to Ascilab to make the change. Admittedly, the standard tuning highlighted the treble detail, potentially leaving a better first impression. And at any rate, it would be impossible to know 100% the correction is needed across all people and all content. But there is enough reasoning here to make a change. As a compromise, Asci chose to use a softer filter than I used.

So here you actually used ears to consider a change. What was the point of that if there's no point in listening?

5. The ship has sailed on measurements are not good enough to tell the story of a speaker. This kind of response is not going to get a pass (TAD in stereophile measurements):

It tells a story, but it's not possible to perfectly balance a speaker just by evaluating the measurements. Or at least, whoever can, should audition for Rainman 2.

 

[amirm]

So here you actually used ears to consider a change. What was the point of that if there's no point in listening?

That was right there in my post. Ability to perform instant AB comparisons, blind and repeated. Listening to a speaker, then tweaking a part, and listening again will not remotely approximate this.

 

[Sigberg Audio]

But don't let that ad-hoc testing lead you to create a messed up response.

But that's not really what happening, is it? I would humbly suggest my designs don't have a messed up response, and generally speaking the Andrew Jones stuff are relatively linear too? This thread isn't about defending Zu Audio.

 

[Sokel]

That was right there in my post. Ability to perform instant AB comparisons, blind and repeated. Listening to a speaker, then tweaking a part, and listening again will not remotely approximate this.

Yes, time is a no-go.
Easy to do with two though, one tweaked, the other not, switch can be instantaneous this way.

 

[amirm]

While I know the story being told on ASR is that ears are completely unreliable, this simply isn't true. Sure there are lots of bias involved, but that's not the same as it being useless. Harman themselves talk about trained and untrained listeners, and have developed software to help you become better at discerning changes in the sound. Does this ability go completely out the window unless you're in a double blind situation? No, it does not.

You jumped on only a part of that quote. I explained that a proper test requires 4 or 5 speakers being randomly compared, and statistical analysis performed to see if the results are repeatable. And listeners can produce useful response. Only then you know if there is a reliable impairment.

Trained listeners with one speaker, listening to just one speaker, cannot produce that result.

 

[Sigberg Audio]

That was right there in my post. Ability to perform instant AB comparisons, blind and repeated. Listening to a speaker, then tweaking a part, and listening again will not remotely approximate this.

I'm probably slow here but I'm not sure I'm connecting the dots.

I can instantly measure, listen and tweak the speaker in any order (active speakers with DSP) - and since I have previous measurements I also know what a tweak will do to the measurements without necessarily re-measuring between every single individual tweak.

How is that different from what you did?

 

[Sigberg Audio]

The most enlightening aspect of the last 100 or so posts in this thread has been that it has exposed those who believe that Guru Speak is a Higher Truth. That audio science only gets a speaker so far (a kind of robotic, formulaic sound), then the finishing 'human touch' with Guru Polish really transports it to the next level.

To believe so is a delusion. A delusion that the supplier marketplace wants us to have. A delusion, indeed, that many consumers want themselves to have, too. And as Arindal so hastily reminds us, we are free to choose that path. Free as a bird.

But don't come in here and try to sell the magic pudding as real magic, or its chef as a wizard. Not if you don't like reality checks.

cheers

Who is trying to sell magic pudding?

All I can say is that I hope most companies will listen to you guys and only measure and not listen to their products before they go out the door.

 

[amirm]

But that's not really what happening, is it? I would humbly suggest my designs don't have a messed up response, and generally speaking the Andrew Jones stuff are relatively linear too? This thread isn't about defending Zu Audio.

It is not. But that measurement shows errors in response which I would not accept in a proper design, sold at high prices. Clearly a number of issues are at play above 1 kHz for a $5,500 speaker pair. This is the reference I compare it to:

 

[Mnyb]

I think all speaker manufacturers should rely on measurements, but as a final step, they should all hire 79-year-old men to listen to them and make sure they sound good.

That should be a sticker just like “Hirez” “octogenarian optimised”

 

[Sigberg Audio]

You jumped on only a part of that quote. I explained that a proper test requires 4 or 5 speakers being randomly compared, and statistical analysis performed to see if the results are repeatable. And listeners can produce useful response. Only then you know if there is a reliable impairment.

Trained listeners with one speaker, listening to just one speaker, cannot produce that result.

So am I understanding it correctly that in your opinion, when I listen to a speaker during the development process, and alternate between measuring, adjusting and listening, the listening part is completely unreliable.

In that case I don't see much point in arguing further, let us agree to disagree. If I misunderstand, I apologize in advance.

 

[amirm]

I can instantly measure, listen and tweak the speaker in any order (active speakers with DSP) - and since I have previous measurements I also know what a tweak will do to the measurements without necessarily re-measuring between every single individual tweak.

You have an active DSP speaker and propose that we should accept non-flat on-axis response like what I just post???

 

[Mnyb]

Say something about the product when field coils are implanted before designing a fully active speaker.

There is only so much you can do passively. I expect all the world best speakers to be active not passive.

 

[Sigberg Audio]

It is not. But that measurement shows errors in response which I would not accept in a proper design, sold at high prices. Clearly a number of issues are at play above 1 kHz for a $5,500 speaker pair. This is the reference I compare it to:

I know it goes against your world view and I respect your view and opinion, but through my own development I have found the anechoic measurements to be a very valuable and important tool, but not all that is needed.

I also generally don't find these textbook measuring speakers to be the best sounding speakers I've ever heard. So the CTA2034 still doesn't tell the full story. In my opinion.

 

[amirm]

So am I understanding it correctly that in your opinion, when I listen to a speaker during the development process, and alternate between measuring, adjusting and listening, the listening part is completely unreliable.

I don't know about "completely" but with the words as written, yes.

I expect you to optimize the design using measurements. If at the end of the line, you can't quite there without making other compromises such as I listed, then sure, you can make fine tuning. But a design loop that is all by ears, with no controls, no other listeners, no notion of what is a reliable content to use, etc., is of little value. I can't verify any of your effort. Instead, I will measure it and see where are.

As I said, the era or speakers that deviate from flat on-axis is gone in my book. No amount of stomping on the ground that you know how to listen, tune, etc. is going to get me there. Not only do I believe flat on-axis is the ideal response, it also gets us to a standard which we badly need.

 

[Newman]

Harman themselves talk about trained and untrained listeners, and have developed software to help you become better at discerning changes in the sound.

I believe the trained listeners land on the same preferences as untrained, but can do it more efficiently. You seem to think that something different happened.

Does this ability go completely out the window unless you're in a double blind situation? No, it does not.

Why not? Nobody was being trained to subvert their own cognitive biases. You are just making an ambit claim and wishing it were true.

If it were true, then audio researchers the world over only need to train their listeners then immediately save millions of dollars by conducting all future audio research with no listening controls. Hmmmm.

As we've shown even trained listeners are biased by price, brand, design, etc.

QED


cheers

 

[amirm]

I also generally don't find these textbook measuring speakers to be the best sounding speakers I've ever heard.

I am the opposite of you. All of my all time favorite speakers have very near ideal response.

Sure, you can get better dynamics out of a huge speaker with flaws in response. That's not the question. Question is whether that speaker would perform better if it had no flaws.

 

[Sigberg Audio]

You have an active DSP speaker and propose that we should accept non-flat on-axis response like what I just post???

I believe I go further, I actively propose that non-flat on-axis response may sound better. But I am then mainly talking about tonality (slopes), not randomly adding peaks or dips, unless it's to balance out off-axis to provide an even listening window / in-room response.

Here is my most recently iterated product:

Measurements for speaker Sigberg Audio SBS G2

www.spinorama.org

 

[Sigberg Audio]

I believe the trained listeners land on the same preferences as untrained, but can do it more efficiently. You seem to think that something different happened.

Do I?

Why not? Nobody was being trained to subvert their own cognitive biases. You are just making an ambit claim and wishing it were true.

So while I'm tuning the speakers, what cognitive bias would be in play typically?

If it were true, then audio researchers the world over only need to train their listeners then immediately save millions of dollars by conducting all future audio research with no listening controls. Hmmmm.

This feels like a strawman.

 

[Sigberg Audio]

I don't know about "completely" but with the words as written, yes.

I expect you to optimize the design using measurements. If at the end of the line, you can't quite there without making other compromises such as I listed, then sure, you can make fine tuning. But a design loop that is all by ears, with no controls, no other listeners, no notion of what is a reliable content to use, etc., is of little value. I can't verify any of your effort. Instead, I will measure it and see where are.

But I have never claimed to have a design loop that is all by ear without controls or other listeners, quite the contrary.

As I said, the era or speakers that deviate from flat on-axis is gone in my book. No amount of stomping on the ground that you know how to listen, tune, etc. is going to get me there. Not only do I believe flat on-axis is the ideal response, it also gets us to a standard which we badly need.

That is fair.

 

[Marin]

Pjesma koju gotovo nitko ne koristi lol

Woow thank you so much for sharing this song......
Incredible voice, and it goes down in the low 60 Herz... just by looking Spectroid app.
Makes me think how this song sound on speaker that has port tuned in range of 80-100 Herz?
Or port tuned just wrong, just some example...

Edit:I just spent an hour listening to his covers of famous songs and the hour went by in a minute
Dude, if you ever find yourself in Croatia, look for me, you'll get a free round of beer from me