Is Jay talking about Amir?

[Hemi-Demon]

For a while I thought this is not North Korea where dictator decides what data people need to see and what is important or just interesting for them. But of course you have every right to keep others deaf, idiots or minors under your guardianship. I can't tolerate such arrogance so hopefully someone with moderator rights is able to erase me if you can't.



You are talking about single wide band systems such as amplifiers. That kind of simplified assumption is not whole truth with multiway loudspeakers having both mechanical and acoustical links between different ways/bands. Already mentioned few mechanisms such as cone of conventional hifi coaxial (Kef, Seas etc) which acts like wave guide vibrator causing IMD products at HF. That is completely different than result of non-linearities of individual radiators, and can be higher and more audible than harmonics - especially in 2-way applications. PA coaxials usually have some kind of wave guide for tweeter which isolates wavefront of tweeter from vibrating cone so sound clarity could be significantöy better than with hifi coaxial. Another already mentioned mechanism is pressure leak to rear side of the radiator. Leak biases operating point of the motor causing intermodulation products. This happens if designer assumes that tweeter is sealed while it's actually not (enough). This is not all.
Measurement as harmonics is also limited to bandwidth of output port/stage so you cannot measure non-linearities as harmonics in acoustical domain above ca. cone break-up freq. / 2 without multi-tone.

Ok. Your turn to demonstrate that you know at least something about this topic. No problem to accept that you can't make too much noise in the house with multi-tone measurements. Just be honest.

North Korea references, talk about reaching. Why not take one speaker Amir has reviewed, like say a Kef design (since you keep referencing them), and perform a comparative complete review based upon you technical acumen and post it to your blog, forum or just post it here on ASR. If Amir is doing it wrong, show him how to improve. Just stating that he is wrong is no better than all the various youtube subjective reviews Or better yet produce one of your designs, or at least show some of the data that you are alluding to, over and over. It is not fair to take shots left and right, then flip like you are persecuted, and then scream erase me, without any proof from your vantage point. We are all trying to learn, prove you side.

 

[Blaspheme]

Huh? Where? Or you just mean that the distortion level tracks the output level? That's quite logical, and the signal is supposed to represent typical music.

Basically, they display the red curve as "what you're actually getting" as opposed to the input signal (green). You can then "measure" the distance between blue and red. I didn't communicate this very well, though.
You could limit this to mostly flat speakers, too; kind of a "bonus" measurement for stuff that has already low linear distortion.

Anyway, this exchange just helps my point: it (just) needs discussion.

Thanks for linking to S&R, I'm less familiar with non-English publications. Now that we have in-browser translation, that shouldn't be a limitation any more.

Amir makes much of telling a story, but I found the S&R narrative more coherent and less amateurish (using the word in the literal rather than pejorative sense). They aren't doing it for free, of course. With respect to measurement visualisations, they lack the detailed off-axis information provided by the spin, but that aside the overall set is comprehensive and not excessive in number. I like the level, IMD and spectral graphics, which are less familiar. I'll read a few more examples and reflect.

 

[Sancus]

Always found it strange that they measure at different levels everytime; they should just do 85 dB SPL average at the listening position with 1 m for nearfield speakers, 2 for midfield and 3 or 4 for farfield.

The S&R IMD test is really hard to understand, and not fully explained anywhere that I've seen, only partially. What they're doing is raising the level until a threshold is met. That threshold is based on total distortion in some way.

I tried to break down what they explain about the test here, but the KH420 review also used an older version of it that is again different. They've changed it a couple times.

The great difficulty in not just interpreting those graphs, but attaching them to any audible meaning, does tend to indicate to me that they're not all that useful. I appreciate that they're there but... I absolutely see the side of that argument that says you should not include graphs that are exceptionally difficult to interpret and which will almost never contain much more information than "3-way better than 2-way, duh".

The problem clearly exists with 2-ways, the old Neumann document seems clear about that. Whether it is worse with 2-way coaxials or not is unknown and I have not seen any source which indicates a clear and consistent way of measuring and displaying this so that it's both comparable and readable across all types of speakers.

Neumann IMD document

Spoiler

 

[Blaspheme]

The S&R IMD test is really hard to understand, and not fully explained anywhere that I've seen, only partially. What they're doing is raising the level until a threshold is met. That threshold is based on total distortion in some way.

I tried to break down what they explain about the test here, but the KH420 review also used an older version of it that is again different. They've changed it a couple times.

The great difficulty in not just interpreting those graphs, but attaching them to any audible meaning, does tend to indicate to me that they're not all that useful. I appreciate that they're there but... I absolutely see the side of that argument that says you should not include graphs that are exceptionally difficult to interpret and which will almost never contain much more information than "3-way better than 2-way, duh".

The problem clearly exists with 2-ways, the old Neumann document seems clear about that. Whether it is worse with 2-way coaxials or not is unknown and I have not seen any source which indicates a clear and consistent way of measuring and displaying this so that it's both comparable and readable across all types of speakers.

Neumann IMD document

Spoiler

View attachment 141715

I agree in part. I've looked at the Kii Three, Genelec 8361 and FocalTrio11 reviews so far. The IMD behaviour is quite different. There are differences in the smoothness etc of the spectral as well. I'd certainly benefit from more info on their IMD method and some guidance to interpretation. But I tend to assume the opposite, these visualisations may answer questions about sonic character and performance that FR and spins aren't telling us.

Edit: I haven't read what you've said in the other thread yet, I'll do that now.

Ok, after a quick read, you've decoded the IMD graph top coloured lines, thank you that helps. My guess is the blue line is TD (or THD?) and the hash is IMD, but confirmation would help there. Basically we want those to be low. The different scales and test signal levels in different reviews over time are confounding wrt direct comparisons, obviously. A-weighting is interesting, I think it's quite valid for the treble, and better reflect ts real-world performance, but as I like electronic music, I prefer not to do that for bass (there's probably a better weighting, albeit less universally accepted).

Publications that have print editions or originated in print tend to do smaller graphs (Stereophile is another example). Those originating more recently or as blogs are often better (like ASR, annotation style notwithstanding). I'd prefer to see say vector pdf used for multi-purpose publication as it scales (but depends on instruments outputting vector not raster). Given that, those IMDs would more comparable showing 0-100 dB constantly, with consistent test signal levels. Where to set the distortion threshold is a bit more complex, as is measuring distance.

I continue to like their spectral graphs, easier to read and compare than CSD waterfall. Someone could probably do an algorithm that calculated deviation from ideal via pixel values (or underlying data) and generate a number (or some numbers) for non-visual types.

 

[amirm]

Amir makes much of telling a story, but I found the S&R narrative more coherent and less amateurish (using the word in the literal rather than pejorative sense).

Give me some examples of the coherence you are talking about. And what makes them less amateurish. I am hoping you are not going by word count and product descriptions....

 

[Rottmannash]

Honest question: Who is Jay?

(Asking before I decide whether to watch the video)

Canadian dude who has a YT channel who apparently worked at a hi fi store years ago...pleasant but vapid.

 

[richard12511]

I haven't seen anything particularly rude in kimmosto's posts

"Your output does not indicate much logic. Try to decide which one is correct or clarify your writing."

"Research and reality you refer does not cover special gradient radiators and/or complex listening environments so I think that you know nothing...not much about combination of conditions and possibilities available for me in our house - and many others"

"Your reality is probably as narrow as your output and part of science you have picked for ASR members."

"Anyway, I've had enough of this forum thanks to you. Please erase me."

How is that not considered rude? I think you may be glossing over his tone a bit because you agree with many of his points . I agree with some of his points too, but I'm also a fan of us being kinder to each other. Most of us are here because it's something we enjoy. I think the anonymity and indirectness of the internet(even if one's real name is known) makes us meaner to each other than we would otherwise be in person.

His more recent posts have been much kinder. We can all understand online frustration. I know I certainly can.

 

[richard12511]

Unless I am mistaken, he is the one who design taipuu speaker (https://taipuuspeakers.fi/en/taipuu-speakers-dsp-active-speakers/).
I am not a expert and I am not able to judge the quality of its work but from my perspective its work (VituixCAD & taipuu) indicate that he has more knowlegde and experience in speaker design than 90% of the members here.

If those are his speakers, they do look pretty damn good. All the measurements he gives look great, looks like they should be very dynamic, and the speakers themselves are quite nice to look at . Would love to see more data.

 

[redshift]

The graphs depicted in this post are pretty ordinary, as we've discussed, so doing more like that won't help much. Your waterfalls also. When you get better at visualisation, maybe. Have you read Tufte?

Every data analyst professional uses Jupyter notebook, pandas and Python.

 

[Sancus]

The graphs depicted in this post are pretty ordinary, as we've discussed, so doing more like that won't help much. Your waterfalls also. When you get better at visualisation, maybe. Have you read Tufte?

Visualization isn't really the problem with THD/IMD graphs. The problem is that those measures are inherently low value, because they don't correlate in an obvious way to anything that is actually audible.

There are various papers out there that rethink distortion as a measure that is directly related to actually audible phenomena, like the Geddes ones, but there doesn't seem to be anything that has really caught on.

 

[KSTR]

How is that not considered rude? I think you may be glossing over his tone a bit because you agree with many of his points . I agree with some of his points too, but I'm also a fan of us being kinder to each other.

This is not rude in my book, just a bit provocative but basically correct, both in style and content. And yes, if somebody is a heavyweight I'm more inclinded to tolerate glitches. It's all about the message, not the messenger.

 

[redshift]

I dunno, but to Swedes Norway and Finland have only temporary escaped them, but they'll come back some day and own us! Or not So Swedes "think" they've invented everything, or something to that effect. The realities is that once upon a time Norway offered the Swedes 10% of the North Sea oil and gas for 10% in Volvo, but in the end the deal was broken by the Swedes. Thankfully! Dumb Swedes

Right, we’re experts at being best buddies and pretend we’re not for the drama and comedy of it.

Q: What's the difference between Swedes and mosquitoes?
A: Mosquitoes are only annoying in the summer.

P.S. I’m a Swede.

 

[redshift]

"Your output does not indicate much logic. Try to decide which one is correct or clarify your writing."

"Research and reality you refer does not cover special gradient radiators and/or complex listening environments so I think that you know nothing...not much about combination of conditions and possibilities available for me in our house - and many others"

"Your reality is probably as narrow as your output and part of science you have picked for ASR members."

"Anyway, I've had enough of this forum thanks to you. Please erase me."

How is that not considered rude? I think you may be glossing over his tone a bit because you agree with many of his points . I agree with some of his points too, but I'm also a fan of us being kinder to each other. Most of us are here because it's something we enjoy. I think the anonymity and indirectness of the internet(even if one's real name is known) makes us meaner to each other than we would otherwise be in person.

His more recent posts have been much kinder. We can all understand online frustration. I know I certainly can.

In all fairness, Finns are well known for not sugarcoating how they feel about stuff.

I find it hilarious.

 

[boselover61]

In all fairness, Finns are well known for not sugarcoating how they feel about stuff.

I find it hilarious.

I hope all this Finns talk are sarcastic. I honestly can't tell.

 

[redshift]

I hope all this Finns talk are sarcastic. I honestly can't tell.

Get to know one. You’ll soon enough understand.

They’ll make the man Jesus himself look like an amateur.

 

[amirm]

This is not rude in my book, just a bit provocative but basically correct, both in style and content.

You need to throw your book away then. Were you raised to go into someone's home for dinner and spit on their food because you don't like it? Correct in style? Good grief. Have I let the boundaries of this forum be this soft for you to be so emboldened?

 

[redshift]

You need to throw your book away then. Were you raised to go into someone's home for dinner and spit on their food because you don't like it? Correct in style? Good grief. Have I let the boundaries of this forum be this soft for you to be so emboldened?

1000’s of people enjoying the objectivism and filling up the review threads. I doubt there is much spitting on average, rather infinitesimal.

However, talking the differential of a signal might expose further avenues of exploration however ill behaved and defined it is.

 

[ctrl]

The amplitudes of the multitone are clearly modulated by the frequency response of the speaker. Where are you seeing the normalization or percentage?

Normalization (to the frequency response) is not allowed in this case because the measured distortion values behave in a non-linear way and thus normalized curves would not be comparable.

But I have to agree with you, the presentation of IMD of S&R (sound and recording magazin) is not helpful in terms of comparability with other speakers.
In the PA market it might be helpful to know that for a loudspeaker A the average IMD of 10% is reached at 104dB SPL and for loudspeaker B at 107dB.

It would be much better if there were IMD measurements for fixed sound pressure levels (e.g. 85, 95, 105 dB) and then the limits for 1%, 3% and 10% IMD, based on the individual frequency response of the loudspeaker, were plotted there.

This could then look something like this:

Source: https://www.lowbeats.de - additional curves were drawn by me
In this way, the measured multitone distortions MD (HD + IMD) are directly comparable. One could also immediately see where a speaker has particularly big problems.
For me this kind of evaluation is easy to understand, but I don't know anyone who uses this (apart from me ).

It would be even easier to understand if the MD were given directly as a frequency-dependent percentage value curve - I agree with Amir on this.

Whether this kind of evaluation of IMD brings an added value or whether the measured harmonic distortions basically already anticipate the result of the IMD measurement in 99% of the cases***, would have to be examined more closely on the basis of selected loudspeaker measurements.

*** Update:
The statement refers to "comparable" speakers. If one 2-way loudspeaker has an f3 of 80Hz and another has an f3 of 40, there are of course significant differences in IMD (for the same driver size) that are not captured by an HD measurement.
In this case, the IMD measurement provides added value.

If both speakers were equalized to the same frequency response, then most likely the speaker with the lower HD will also have the lower IMD (but that is just my experience).

Visualization isn't really the problem with THD/IMD graphs. The problem is that those measures are inherently low value, because they don't correlate in an obvious way to anything that is actually audible.

I couldn't agree more. The THD evaluation of S&R does not allow a comparison of the results of the individual tested speakers, because it is not clear by which order of harmonic distortion the THD is significantly influenced.

The measurements of the harmonic distortions HD, splitted into HD2 to HD5, as shown by @amirm or @hardisj, are much more informative, especially because the masking decreases significantly with the increase of the order of the HD.

There are various papers out there that rethink distortion as a measure that is directly related to actually audible phenomena, like the Geddes ones, but there doesn't seem to be anything that has really caught on.

The Geddes and Lee studies were certainly in the right direction, but unfortunately the correlation was not very high.

In listening tests, the authors found a moderate correlation between this metric and the subjective impres-
sions of artificially applied distortion on the magnitude of 0.67 [9]. The GedLee metric is not applicable
to nonlinear systems which are frequency dependent since the metric assumes that the transfer function,
21T (x), is valid for all frequencies. The nonlinear distortion in real transducers often varies with fre-
quency. As such, it was recommended in [26] that the GedLee metric be extended into frequency bands
and applying the metric separately in each band.

Source: Perception & Thresholds of Nonlinear Distortion using Complex Signals

There are other approaches that are more promising, but for many completely unknown - see linked study.
The study is already a few years old, I am not up to date on the latest developments.

 

[q3cpma]

The S&R IMD test is really hard to understand, and not fully explained anywhere that I've seen, only partially. What they're doing is raising the level until a threshold is met. That threshold is based on total distortion in some way.

I tried to break down what they explain about the test here, but the KH420 review also used an older version of it that is again different. They've changed it a couple times.

The great difficulty in not just interpreting those graphs, but attaching them to any audible meaning, does tend to indicate to me that they're not all that useful. I appreciate that they're there but... I absolutely see the side of that argument that says you should not include graphs that are exceptionally difficult to interpret and which will almost never contain much more information than "3-way better than 2-way, duh".

The problem clearly exists with 2-ways, the old Neumann document seems clear about that. Whether it is worse with 2-way coaxials or not is unknown and I have not seen any source which indicates a clear and consistent way of measuring and displaying this so that it's both comparable and readable across all types of speakers.

Neumann IMD document

Spoiler

View attachment 141715

Found the document here: Glossary for studio monitors: Measurement descriptions
Neumann also has a ton of cool PDFs from their K+H days. I warmly recommend FAQ for studio monitors: General product answers too for most people, their explanations are clear and useful.

In any case of "we don't know", the "lower is better" rule is always good, unless it means a compromise elsewhere.

 

[Koeitje]

So what is exactly the fight here? Is it that @kimmosto doesn't think CEA 2034-A-2015 is predictive in terms of loudspeaker preference?