Keeping this on topic about speaker isolation, you have failed to explain why standard REW measurements won't show a difference that can be heard. I proved that your example that AM/FM sound different but measure the same, yet you never acknowledged that. You never acknowledged any of my other points that relate to hearing speaker isolation. You called me wrong a bunch of times, and said my tests are not valid, but you never once explained why they're not valid. At least not in a way that makes sense. If you truly believe you can hear artifacts and differences 90 dB below the music, that tells me for certain that you have no idea what you're talking about. Here's proof of that for you to ponder:
http://ethanwiner.com/audibility.html
http://ethanwiner.com/audibility.html
True. Nor does whether or not I can pass your (possibly non-scientific) listening tests. Yet you attempt to goad me into playing your game. Why do I say that? I know quite well it is possible to do things like make hi-res sound exactly like CD. Its also possible to make hi-res sound better than CD. Probably it could be made to sound worse too. So, I return to the message in #227: there are thousands of ways to design testing the wrong way, and only a few ways to do it right. Based on everything I know about you so far, I am not convinced you are the equivalent of a competent research scientist. You may intend to be perfectly honest with your listening tests, but as a statistical likelihood you may have discovered some of the thousands of ways of designing bad experiments. So, is it on topic for me to visit your website and then for us to talk about that or not?
^^^ Utterly bizarre. You proved nothing about AM/FM; you missed the point completely about placing too much faith in spectral analysis, and you didn't figure out a way to make a proper FFT that showed the individual frequencies. You also still don't seem to understand a word about what REW is good for and what it isn't so good for. And I know at least some people understood it:
I have some idea of who @mark100 might be referring to with respect to fixation on frequency domain response. You keep going back to REW without ever acknowledging you understand what REW doesn't show about information contained in spectral line skirts. You still seem to have insufficient understanding of human perception of phase and the lack of information about in FR spectral measurements.
Moreover, its seems like you already forgot about ITD and the fact that you can't measure stereo effects of isolators with only one speaker and one mic.
Sorry if that stuff wasn't plain enough or if it didn't stick with you. All I can say is I am willing to try to explain again if you really want to know how much you are missing.
Last edited:
OMG, you are so full of it. You have no interest in learning, you just want to pretend you already know everything. So forget the HD versus CD test and just try the other two. But you won't because deep down you know that you won't pass, and you'll be embarrassed when I tell you how well you did. So I really am done with you. Let me know when the AES invites you to present two workshops, and when you write an 800+ page book about audio that's used as the main course text at major universities like Notre Dame and Tufts. Sorry if I wasted your time. 😒
https://ethanwiner.com/book.htm
Ethan, first of all I don't know it all. Nobody does so far as I can tell. Audio is a huge field, much like electronics itself. Regarding your book, do you really want to open that off-topic can of worms? I have seen the book. A lot of it is good and correct. Kudos for that. That's the good news. There will have to be some bad news too if you really want to litigate it here.
Regarding learning, you don't seem to have learned anything in this thread, not from gedlee nor from anyone else. OTOH, I do pay attention. Gedlee said that imaging and localization are treated separately in the literature. https://www.diyaudio.com/community/...eaker-cabinets-decoupling.428300/post-8037712 So, I will keep that in mind from now on. You OTOH, don't seem to have learned anything. But if I am wrong and you learned something technical that you will now keep in mind forever, what was it?
Last edited:
Gedlee,
I would like to investigate this.
Just my opinions before further research.
Hard Rubber? Transmits Vibrations. Yes?
Soft Rubber? Transmits less Vibrations. But wears out, gets punctured by the metal post.
Sorbethane? Transmits less Vibrations. Might have to be replied after every song.
I would like to investigate this.
Just my opinions before further research.
Hard Rubber? Transmits Vibrations. Yes?
Soft Rubber? Transmits less Vibrations. But wears out, gets punctured by the metal post.
Sorbethane? Transmits less Vibrations. Might have to be replied after every song.
Also potentially allows more of the energy to act on the box alone.
The main difference between rubber and sorbothane would have to be their damping qualities.
Hello, this is "i have a problem i solve it" man.
Why would you choose any material when you don't know how, where and why ?
A technical solution is like an answer to a question and if there is no question, there is no answer, so what is the problem?
Why would you choose any material when you don't know how, where and why ?
A technical solution is like an answer to a question and if there is no question, there is no answer, so what is the problem?
We encourage frank discussions about technical issues but if you feel the urge to get personal, here's an effective method to refrain from it.
Move your head slowly 45 degrees to the left, then 45 degrees to the right.
Repeat five or more times until the urge disappears.
We do have an ignore button in case the above doesn't help.
Remember, when the mod team needs to step in, we've got a whole arsenal of unpleasant tools if self-help fails.
An unpersonal and technical disucssion about what ?
Human technology is human as well, they can't be separated unless you are dead and don't need technology anymore.
I got no problem with those who try to solve a problem they don't have with people they don't know but a practical problem is the only thing that interest people in order to start learning something, whatever it is.
Human technology is human as well, they can't be separated unless you are dead and don't need technology anymore.
I got no problem with those who try to solve a problem they don't have with people they don't know but a practical problem is the only thing that interest people in order to start learning something, whatever it is.
Other than that, all the system needs to be evaluated with the ears.
A competent ( no other adjective comes to mind) source, a competent ER...piece of music - what about 'inaging ' an orchestra? Too 'hot' for
Your system? My problem too, but go ahead..
So a competent volume
A competent amplifier....
A competent ( no other adjective comes to mind) source, a competent ER...piece of music - what about 'inaging ' an orchestra? Too 'hot' for
Your system? My problem too, but go ahead..
So a competent volume
A competent amplifier....
Paraphrase: “everything is a compromise.
I have mounted subs inthe ceiling (of a club).
And then he talks about about something like AllenB’s spring/damper system.
dave
I just got back from a 2 week trip to lovely Amsterdam and Norway. Unlike Ethan, I am 75 and travel extensively.
I glossed over the recent comments and was dismayed by the level disinformation presented. I will not go over specific posts as that would be too time consuming.
As to the idea that we can't measure everything that we can hear, that is simply false. Everything that you hear can be measured both linear and nonlinear, but we may not as yet actually measure everything. That said, Toole and colleagues have shown that their measurements account for 95% of the variance in the blind tests that they do. That means that, at best, what we aren't measuring is in the 5% range. However that 5% could also just be statistical noise in the experimental design and not anything of audible significance. Because 5% is so small it is a significant task to prove that what you think that you hear is actually there. Significant claims require significant proof.
Notice that I said "hear", there are tactile issues that are never discussed or measured that can be important. I've alluded to some of those before.
I won't reflect on "errors" in electronics simply because I have never seen any proof that such errors are audible in a well designed circuit. Hence they aren't worth worrying about. Sure there might be some pieces with audible "errors" but these should be the rare exception, and are certainly not common.
FM and AM of a pure tone will have identical levels of sidebands. But they sound different, how is that possible? That's because the sidebands have opposite phases for the two modulations and this would only be seen if one looked at the phase, which a spectral plot does not usually show - it's not shown because it lacks significant importance (there are some issues to be sure, but it's mostly the group delay rather than the actual phase,) but this phase information is there in the data.
I saw this clearly when I did an IMD measurement on a full range driver. (In my youth I believed that nonlinearities in loudspeakers were important. Toole also doesn't give them much importance. Maybe in the 5%?) The sidebands were not symmetric! What I was measuring was both the AM and doppler distortion where the sidebands canceled on one side and added on the other.
I agree with almost everything that Ethan has claimed, although his explanation of the physics of sound transmission through isolators was wrong. I agree, "There is no magic in audio." As someone here said, it's difficult for the novice to parse out the good info from the bad. That's why I try and correct misinformation when I see it. Of course you have to trust that I know what I am talking about - few have my bonafides.
Markw4: I don't think that you understand spectral analysis as well as you think that you do. For example you said:
"(1) FFTs have an averaging effect, the larger the FFT size (i.e the finer the frequency resolution) the more its giving an average measurement." is completely false. Either you misstated it, or I misunderstand what you are trying to say. The larger the FFT the less averaging effect there will be. For example, a 2 point FFT averages the spectrum into two bands, a larger FFT averages the spectrum only over each bin in the FFT, i.e. far far far less averaging as the FFT size gets larger. Then there is the discussion of using steady state signals, etc. Those were also misleading. Only steady state measurements make any sense for characterizing a system.
Don't even get me started on nonlinearities, both subjectively as well as how they are measured.
I glossed over the recent comments and was dismayed by the level disinformation presented. I will not go over specific posts as that would be too time consuming.
As to the idea that we can't measure everything that we can hear, that is simply false. Everything that you hear can be measured both linear and nonlinear, but we may not as yet actually measure everything. That said, Toole and colleagues have shown that their measurements account for 95% of the variance in the blind tests that they do. That means that, at best, what we aren't measuring is in the 5% range. However that 5% could also just be statistical noise in the experimental design and not anything of audible significance. Because 5% is so small it is a significant task to prove that what you think that you hear is actually there. Significant claims require significant proof.
Notice that I said "hear", there are tactile issues that are never discussed or measured that can be important. I've alluded to some of those before.
I won't reflect on "errors" in electronics simply because I have never seen any proof that such errors are audible in a well designed circuit. Hence they aren't worth worrying about. Sure there might be some pieces with audible "errors" but these should be the rare exception, and are certainly not common.
FM and AM of a pure tone will have identical levels of sidebands. But they sound different, how is that possible? That's because the sidebands have opposite phases for the two modulations and this would only be seen if one looked at the phase, which a spectral plot does not usually show - it's not shown because it lacks significant importance (there are some issues to be sure, but it's mostly the group delay rather than the actual phase,) but this phase information is there in the data.
I saw this clearly when I did an IMD measurement on a full range driver. (In my youth I believed that nonlinearities in loudspeakers were important. Toole also doesn't give them much importance. Maybe in the 5%?) The sidebands were not symmetric! What I was measuring was both the AM and doppler distortion where the sidebands canceled on one side and added on the other.
I agree with almost everything that Ethan has claimed, although his explanation of the physics of sound transmission through isolators was wrong. I agree, "There is no magic in audio." As someone here said, it's difficult for the novice to parse out the good info from the bad. That's why I try and correct misinformation when I see it. Of course you have to trust that I know what I am talking about - few have my bonafides.
Markw4: I don't think that you understand spectral analysis as well as you think that you do. For example you said:
"(1) FFTs have an averaging effect, the larger the FFT size (i.e the finer the frequency resolution) the more its giving an average measurement." is completely false. Either you misstated it, or I misunderstand what you are trying to say. The larger the FFT the less averaging effect there will be. For example, a 2 point FFT averages the spectrum into two bands, a larger FFT averages the spectrum only over each bin in the FFT, i.e. far far far less averaging as the FFT size gets larger. Then there is the discussion of using steady state signals, etc. Those were also misleading. Only steady state measurements make any sense for characterizing a system.
Don't even get me started on nonlinearities, both subjectively as well as how they are measured.
The hardness of the rubber is not really a significant factor in transmission, it's more about the damping. The hardness is usually determined by the static load and the allowed static deflection. Once the system has settled statically the compliance of the system is about the same.
I am not an expert on commercial products although there is a huge amount of work done on this for all areas of transportation. Look at the damping of the isolator more than its compliance.
True.
Have you ever measured audibility of 1/f-like current noise which has been intermodulated with the audio signal? If not, IME in the case of a FIRDAC where the output array resistors have high current noise, it can be easily audible by anyone at LF, but only if listeners are trained to know what to listen for. Otherwise they may not know what it is they are hearing. They may think its part of what was recorded in a live performance.
Of course, the problem with it is that its is both a distortion and a noise, but it sounds like neither.
Short time FFTs are used to measure nonlinear systems because they have less averaging effects of time-domain changes. They are used for things like EKGs, EEGs, etc. The signals are not only non-LTI, they are also non-stationary.
Its simply a consequence of the well known fact that more resolution in the frequency domain corresponds with less resolution in the time domain. Thus, signals which are changing in time show up as more averaged effects in high resolution FFTs.
Another way of explaining the effect is that DFTs can be calculated by correlation. That's one of the old ways of calculating them by hand. A non-PSS signal, one that is changing frequency, amplitude, being gated on and off, etc., during the time an time-series dataset is being acquired for FFT analysis may have low correlation on average with a particular FFT bin. An easy to imagine but unlikely example would be a frequency sweep across the audio band while FFT data is being acquired. If equal power of the sine sweep is correlated with each bin frequency, then the sine sweep will look like a noise floor. Obviously that's not what it really is.
Last edited:
And this has been shown to be statistically audible in a blind test were? If not then I'm not interested.
Again, electronics is not my expertise, but I would be shocked if the level of noise in a circuit was high enough to modulate the signal. In fact, if the system is linear (which I hope it would be) then "modulations" are not possible because only a nonlinearity could do this. The noise and signal are uncorrelated and can only interact with each other through a nonlinearity. This is fundamental systems theory.