The entire transfer function is a chain from microphone to ear, and there is nothing to prevent maximising the fidelity of the transfer function at each point in the chain. If we concentrate on one particular link, we find the law of diminishing returns blunts our successes somewhat. However, loudspeakers (and the rooms in which they operate) remain weak points in the chain regardless of how the recordings have been processed. Hence improving the accuracy of loudspeakers is (for some) a worthwhile endeavour.
However, just because the fidelity of the transfer function can be quantified completely does not mean the artistry in recorded music is destroyed in any way. Like you I suspect, I can take apart the laptop I am using now, measure precisely the signals flowing around the motherboard, and then work out where a problem might be causing an error in one of those signals. Critically we can manage that task without the slightest idea of the artistic content in those signals, and we can rest assured there will be a greater chance that the artistry has been better preserved by correcting the error.
The artistry apparent in the content is not then part of the discussion in this thread, since, as you state, such a discussion would struggle with validity. This discussion here instead concerns the best preservation of artistic content in the path from the microphone(s) to the eventual ear(s). And in that endeavour, there is no issue concerning validity, and we can even substitute test signals for artistic content if we like to make our job easier. We are not concerned with the information content, only the accuracy of its transfer.
Nothing is likely to be wrong with it (one would hope). But I don't think either that it lays claim to provide a complete characterization of loudspeaker drivers or of the subjective differences between them.
I do not have access to that particular standard, but as an example raised in this thread, does it establish a basis for separating displacement-dependent non-linearities from current-dependent non-linearities by applying some weighting that models their different subjective thresholds?
Last edited:
Hi mayhem13,
I agree with your points about dehumanizing through tech. But this is again a balance. We have to recognise when tech is the way to go, keeping our common sense and intuitive knowledge engaged is also important. It is also far too common for the science to be ignored, and that never ends well. When I make equipment improvements, I apply knowledge gained through my test and measurement background and measurements of both components and the system I am improving. This way improvements are real and often audible. I couldn't do this without that base in knowledge and testing capability. This is pure science at work, maybe someone could argue it is art, but I'm just following rules.
I sold most of those speakers you're talking about in those times. I also dealt with ancient horn loaded systems and even know a friend with EV Patricians in service in his home. Another good friend has had many very high end horn loaded systems, classic ones. He finally found happiness last Christmas with a system that delivers everything (and I agree, wish I could afford them). The Klipsch Jubilee speaker system. Many are memorable, like my favorite from the time, the AMT 1a with the Heil air motion tweeters. But I have to say, as good as they were, old technology is often remembered in context to their contemporaries. Compare them to today's equipment and they do fall short. The Marantz 10B tuner isn't as good as my Revox 260s or 261 tuners. The AMT 1a, I'd rather have my PSB Stratus Gold speakers, or a new set of Klipsch (RP-7III would be nice).
So while we could do pretty well back in time, improvements in design based on measurements (when we get those right) and materials made products better. It takes a human element to make decisions and balance things. But we simply can't get this far without having metrics to guide us. Those metrics come about through test and measurement. We do check to see "how it sounds" as a sanity check.
I agree with your points about dehumanizing through tech. But this is again a balance. We have to recognise when tech is the way to go, keeping our common sense and intuitive knowledge engaged is also important. It is also far too common for the science to be ignored, and that never ends well. When I make equipment improvements, I apply knowledge gained through my test and measurement background and measurements of both components and the system I am improving. This way improvements are real and often audible. I couldn't do this without that base in knowledge and testing capability. This is pure science at work, maybe someone could argue it is art, but I'm just following rules.
I sold most of those speakers you're talking about in those times. I also dealt with ancient horn loaded systems and even know a friend with EV Patricians in service in his home. Another good friend has had many very high end horn loaded systems, classic ones. He finally found happiness last Christmas with a system that delivers everything (and I agree, wish I could afford them). The Klipsch Jubilee speaker system. Many are memorable, like my favorite from the time, the AMT 1a with the Heil air motion tweeters. But I have to say, as good as they were, old technology is often remembered in context to their contemporaries. Compare them to today's equipment and they do fall short. The Marantz 10B tuner isn't as good as my Revox 260s or 261 tuners. The AMT 1a, I'd rather have my PSB Stratus Gold speakers, or a new set of Klipsch (RP-7III would be nice).
So while we could do pretty well back in time, improvements in design based on measurements (when we get those right) and materials made products better. It takes a human element to make decisions and balance things. But we simply can't get this far without having metrics to guide us. Those metrics come about through test and measurement. We do check to see "how it sounds" as a sanity check.
I'd love to hear/learn more!Actually, in DBLTs, electrostats don't do as well as some well designed electromagnetic drivers ... and some using Unique Fibrous Material (UFM aka paper) do a LOT better than Beryllium and other fancy diaphragm materialsPoorer 'waterfalls' for da ELS units point to some of the reasons why.
aaaagh i wish you would have left out the PSBs......the midrange driver breakup on those is horrible and.....well....plain to see.........i'll stop there. LOL
I am not an expert, but may I add my 2c?
1. standards shall be agreed between vendors and users, like in ITU-T. here I see only the vendors' side. nothing from users.
2. the language, approach, and methods are stone-age archaic laughing stock "The measurement resolution shall be no wider than 1/20 octave"
3. "All testing shall be conducted in an anechoic chamber or equivalent environment". As a user, I can't care less how a speaker performs in an anechoic chamber. I need it to sound well in my room, when placed by the wall, in the wall, at the corner, and in the corner. "It sounds so well in a $1,000,000 anechoic chamber" does not count as an answer.
4. These guys simply have no idea of DSP and the Theory of System Identification. They think in terms of measurements from the 70s, pre-DSP, pre high-res codecs, not in terms of full-band RIR. By regularised IR inversion, you can easily make any speaker perfectly flat (as Neumann). So what?
5. They simply don't understand the nature of non-linear distortions in loudspeakers.
6. Barkhausen noise is not mentioned.
7. Too heavily biased by cinema and home theater folks. 75 dBSPL is not very quiet, it is plenty loud. You don't have to run your system loud if it sounds very good. BTW, the normal level of undistorted human conversation is 60 dBSPL. The more distorted FE signal is, the louder it must be to be on par with intelligibility.
8. The median distance from viewers to TV in USA and Europe is a bit shy of 10'. 8 ... 12' covers > 90% of use cases. Loudspeakers are placed by the TV side, as close to the walls as possible. 4m is nonsense.
At the times, the "wall of sound" technique was developed to combat obscene non-linear distortions in target users' loudspeakers. The performers and music producers did not care so much if the FR was perfectly flat. These guys are obsessed with things that do not matter.
Great example of how sighted measurements can lead one astray......if you like them.....then that's all that counts.....i wouldn't measure them if i were you......expectation bias is a real b!tch. LOL
Expectation bias is the biggest problem when it comes to assessing anything by ear, or even taste.
The second series of PSB Stratus Gold speakers had a nasty midrange thing going on, I was thinking of "upgrading" to the new speakers. It was clearly not a good direction to go in. So I stuck with the originals. Anyway, maybe those are what you are referring to. Doesn't matter, these are I think mid 1980's design and I'm sure we have better today.
For personal equipment I intend to enjoy, I only measure the electronics. Never speakers, I know I won't like some aspect of them. Better I not know. By the same token, I never assess my own work. I have people who know that what I have them listen to may be original, fixed but unchanged or upgraded. Since they never know, they got wise and report truthfully. Many people want to please and may give a glowing report to something no matter how it really sounds to them. I really value input from kids or women who aren't interested in audio, but like music.
The second series of PSB Stratus Gold speakers had a nasty midrange thing going on, I was thinking of "upgrading" to the new speakers. It was clearly not a good direction to go in. So I stuck with the originals. Anyway, maybe those are what you are referring to. Doesn't matter, these are I think mid 1980's design and I'm sure we have better today.
For personal equipment I intend to enjoy, I only measure the electronics. Never speakers, I know I won't like some aspect of them. Better I not know. By the same token, I never assess my own work. I have people who know that what I have them listen to may be original, fixed but unchanged or upgraded. Since they never know, they got wise and report truthfully. Many people want to please and may give a glowing report to something no matter how it really sounds to them. I really value input from kids or women who aren't interested in audio, but like music.
I’ve got a set of DCM Time Windows and Magnepan LRS plus…..both measure horribly……and bring a huge smile to my face when listening……….both highly regarded by reviewers and publications and both huge sellers in their market…..yet over at ASR, the LRS is the antichrist of audio after a whirl through a members spin o Rama thingy. Best was the picture of the setup with the LRS right up against the wall…….really fellas?
Maybe the guy over at GR Research can fix em with his microphone and premium parts? Lol
Maybe the guy over at GR Research can fix em with his microphone and premium parts? Lol
lol!
I read a tiny bit over at ASR. Not impressed one bit.
I was never a fan of the Maggies, but in the right room positioned properly they probably sound good. The Time Windows were more my speed.
If you can find a pair of AD0163T8, they will sound much better. Drop in replacement for the AD0162T8. The difference is the 0163 loses that huge peak between about 12 KHz and 15 KHz while extending the HF response (they are much flatter). I could look them up in my old data books to give you more exact numbers. That mylar dome in the 0162's made that peak. The Qts is about the same.
So that would be the upgrade. The same holds true for the Dahlquist DQ10, and yank the Piezo (pitching it as far away as possible). Don't gas-bag those.
I read a tiny bit over at ASR. Not impressed one bit.
I was never a fan of the Maggies, but in the right room positioned properly they probably sound good. The Time Windows were more my speed.
If you can find a pair of AD0163T8, they will sound much better. Drop in replacement for the AD0162T8. The difference is the 0163 loses that huge peak between about 12 KHz and 15 KHz while extending the HF response (they are much flatter). I could look them up in my old data books to give you more exact numbers. That mylar dome in the 0162's made that peak. The Qts is about the same.
So that would be the upgrade. The same holds true for the Dahlquist DQ10, and yank the Piezo (pitching it as far away as possible). Don't gas-bag those.
If my single remaining brain cell hasn't gone short circuit from all the above supa dupa stuff, am I to deduce there are NO BISPECTRA MEASUREMENTS OF SPEAKERS OR UNITS on record?
Got links to these papers? (I might have to beg a kind AES member to illegally provide me with copies
) I believe the Wigner Distribution is actually just another fancy 'waterfall' display ... from da previous Millenium when i had more than one working brain cell and cud do sums Gerzon's 'bi-spectral' analysis was to do with theories of localisation. https://secure.aes.org/forum/pubs/conventions/?elib=6827
This important paper encapsulates (gives the same results) as ALL known theories of localisation except for the impulsive HF and pinnae colouration models. Michael used to think I unnerstud him cos I nodded my head at appropriate moments but I was just pre10ding
I did use his theories to design a couple of speakers with improved localisation thoughThat is the case with MLS but NOT with Angelo's log sine sweep. I don't think there is any reason NOT to use Angelo's method these days as he has released it for all and the computing power required is less than for a MLS system.
Last edited:
Not being able to afford such august publications, could you sum this up in words of one or two syllables please.
Pages and pages and I still concur. Is there a dispute about the audibility of nonlinear distortions going on? Since that has been researched over and over.
Are you going to start your own thread?
Dis beach bum is annoyed the weather precludes fishing or diving on da Great Barrier Reef for at least a week. I feel an imminent attack of diarrhea with the urge to pontificate (from probably da wrong orifice) on stuff from my previous life
As a start, can someone tell me which of the free 'acoustic' measurement packages can do
- 'waterfalls'
- use Angelo's method?
Besides none give measurements that I can compare with plots made with my B&K 2307 chart recorder & 2010 (measurements as God intended) by the time honoured method of holding up bits of B&K bogroll to the light. But it's getting increasingly hard to get the steam to run these vintage devices.
My own stuff which replicates these measurements, rely on Windoze XP and XP machines are nearly extinct.
Which ones were big enough to get a listener and speaker(s) in to conduct a listening test?
This isn't a trick question. IIRC, there's a discussion on this and what we had to do to ours, in one of the papers I linked to.
I am not an AES member either, but I will try and dig some out some titles and post them in the next day or two.Got links to these papers? (I might have to beg a kind AES member to illegally provide me with copies
I would not call the Wigner Distribution "just another fancy waterfall display" though. It can be regarded as a superset (the mothership?) of all the waterfall displays for a start. More critically, it contains information that is lost in those others displays, and provides an optimal basis for modelling our auditory capabilities. As I speculated early on in this thread, I think it only failed to catch on because of its double sample rate requirement - an objection that no longer holds much sway.
Michael Gerzon's localization model did indeed include the cross-bispectrum!!Gerzon's 'bi-spectral' analysis was to do with theories of localisation. https://secure.aes.org/forum/pubs/conventions/?elib=6827
This important paper encapsulates (gives the same results) as ALL known theories of localisation except for the impulsive HF and pinnae colouration models. Michael used to think I unnerstud him cos I nodded my head at appropriate moments but I was just pre10ding
Cross refers to the bispectral relationships between the sound pressures incident at the two ears. You will also find references to the bispectrum in and around the extraordinary body of work he published in this period. As you are probably aware, although his metatheory you linked was published in 1992, this was essentially a copy of that he developed in the early 1970s and published as part of the Ambisonics specification (then called "System 45J" I think?). Even now I don't think the magnitude of his work is fully appreciated.It was not my intention to make any comparison, just to give an example of how relevant information is often simply obscured by our analyses. I presume the non-linear data is encoded within the MLS noise, just decoding it is not easy, if indeed it is possible in this case?
I believe so... but I want to let some of the "passion" on this thread burn itself out... otherwise the new thread may just be a repeat of what is going on here...
I remembered the PSBs being reviewed/measured in Stereophile so took a look. They measured pairs in 1991 and 1997:
https://www.stereophile.com/content/psb-stratus-gold-loudspeaker-1991-measurements
https://www.stereophile.com/content/psb-stratus-gold-loudspeaker-1997-measurements