"Things that cannot be measured"

[watchnerd]

Of course the laws of physics dictate that anything that can be sensed aurally can be measured,

Hmmm....

I wouldn't agree.

Placebo effects are sensed.

 

[Lambda]

completely and with no loss of information.

This fft corresponds to what signal?

Its of cause a trick question because they look identical in the frequency domain jet they are clearly different in time domain

 

[KSTR]

There are no "Things that cannot be measured" but can be heard.
But there are Things that cannot be measured with (long)FFT that can be heard.

This is correct and wrong at the same time.
"To measure" is just a way too loosely defined term to allow for any sort of generalization.
It is certainly true that a specific meusurement may fail in peeling out an error mechanism that is really present but then you've just measured the wrong thing with the wrong procedures and may even have a wrong understanding of things. In case of the large size FFT spectrum measurement with a steady-state sine source we know that, for example, thermal settling issues might be easily missed. A Class-A/B amplifier has a quiescent (bias) current setting scheme that has to adapt to the current temperature of the output transistors quickly and without issues of being too slow or to nervous. After high-power passage of music some amps go underbiased and then signals may get distorted more than normally. Even if you measured the amp correctly by applying a preconditioning noise berore taking the measurement, a long FFT over many seconds of data will "smear" the effect. Other techniques should be used for that kind of issues.

In general, the most basic conceptual measurement is the difference measurement (subtractive analysis), comparing the continuous time signal output of the Device Under Test (DUT) with a known Reference. For a Preamp, the Reference would be a simple cable short between the DAC output and the ADC input of the measurement rig.
Any difference, no matter what it is, is fully captured in the difference with a suitable source signal (which can be anything, no restrictions). There is no way it could ever be otherwise, by sheer definition and by first principles. A diff test can and will reveal everthing and will go way below any meaningful threshold for audibility. The diff test needs some special pre-requisites and is, while conceptually simple, a true challenge in the procedureral and technical details, to make it work reliably. I'm dealing with subtracive analysis for some 10years now and happen to have solved a key issue just in the past weeks so that I now can reliably measure (expose visually and audibly, plus lots of plots and numbers) any signal difference, no matter what it is, down to 130dB (and lower) below the signal which is already 20dB below the analog noise floor of even the best equipment. In other words, way beyond any human hearing thresholds even if we allow for 20dB of headroom for the numbers for, well, "just in case". -130dB is a 1 : 10,000,000 ratio (0.0381mm / 0.0015inch of the height of the Empire State Building in New York).

Bottom line: by first principles, every signal change that may ever cause audible differences can always be reliably measured (quantified in some way) and exposed. Period.

 

[KSTR]

This fft corresponds to what signal?
View attachment 115474

View attachment 115475

It's of cause a trick question because they look identical in the frequency domain jet they are clearly different in time domain

Actually they don't look identical. "Frequency Response" means a FR of magnitude and phase and when you have one FFT block of FR data you can go back to the original time-domain signal by iFFT (with any FFT window implicitely applied to that reconstruction, which is why, for example, FFT/iFFT-based convolvers don't use any window functions).

 

[daftcombo]

Until now, you cannot measure the amount of rationality and good sense these reviews & forum bring to the Hi-Fi world.

 

[drfous]

I have no technical background in any of this. It seems to me we should be able to measure the quality of these devices.

But it just seems to me that measuring a speaker using a frequency sweep doesn't measure everything we ask a speaker to do. It's a physical device playing many frequencies at the same time.

Audi diesels measured well, but in the real world they didn't perform in the same way.

why isn't this a potential problems with the frequency sweep measurements that seem to be the standard today?

TIA!

 

[Lambda]

Actually they don't look identical. "Frequency Response" means a FR of magnitude and phase and when you have one FFT block of FR data you can go back to the original time-domain

Everything you said is true. But i was refereeing to FFT in the way it is presented and used here as just a 2D plot without Complex numbers.

, a long FFT over many seconds of data will "smear" the effect. Other techniques should be used for that kind of issues.

Yes that's my other point why i said "(long)"

This sections have (almost) the same over all signal to noise.
but the one section has it all in one place and therefore it's way more audible.

Again the Mathematics behind FFT worke ...
But this FFT measurements don't capture all this information in the resulting 2D graph's we see

 

[Ron Texas]

IQ can be measured. Stupidity can't be measured.

 

[StefaanE]

I'm in the "things that haven't yet been measured" camp. It find it too smug to claim we know every measurable parameter of things at present.

Oh come off it. If it can be pressed in a slab of vinyl, it can be completely, exhaustively measured.

 

[Mnyb]

The biggest uh-oh in these Yamaha claims is that they "tune" the sound ? when they fully well know that it can be made transparent and probably is tuned or not ? There is nothing to "tune" in the sense of a musical instrument .

 

[Lambda]

No the FFT is simply a mathematical algorithm "appears vs actually" here has no meaning.

The Mathematical concept works if implemented correctly.

But i'm referring to "FFT- analysis in Audio Measurements" in the way they are "actually" executed.
like this:
https://www.audiosciencereview.com/...nderstanding-fft-in-audio-measurements.20814/

 

[Wes]

Certain things are not often measured. And that, not ability to measure, is the real issue.

An example in audio would be the spectrum of distortion products, which are often subsumed under THD. What about even vs. odd? And what about crossover distortion? How much of that can we hear - going from A to B? Or... back down?

For speakers (which are highly likely to be a "limiting factor" on audio fidelity than an amp) what do we know about the audibility of cone breakup?
(The recent thread on speaker breakin should have been on breakup...)

There are similar things in env. chemistry. We can usually measure to ppt but where are the toxicology data? If they exist, they are on single chemicals not on numerous ones in possibly synergistics combination, and are acute studies, not chronic ones.

Many other example exist in various fields.

I skipped a few pages after the personal attacks started so maybe the above was already mentioned.

 

[KSTR]

An example in audio would be the spectrum of distortion products, which are often subsumed under THD. What about even vs. odd? And what about crossover distortion? How much of that can we hear - going from A to B? Or... back down?

I hope I could answer that viable question in my previos posts.
And I completely agree, trying to condense any complex phenomenon into a single scalar value like "SINAD @ 1kHz, near full-scale signal" is not that much useful, often. It still does tell something, and that is when the value reaches "astromonical" values as in amps like the Benchmark ABH-2 is very unlikely that such an amp fails severly (audibly) in other regards. A set of proper difference test will tell exactly if there is anything popping up reaching only a measily -50dB or so.

 

[Sgt. Ear Ache]

as has been mentioned several times already, there is a way to prove the existence of these unmeasurable manifestations of "musicality." Until someone does that, it all amounts to wishful thinking.

 

[KSTR]

The biggest uh-oh in these Yamaha claims is that they "tune" the sound ? when they fully well know that it can be made transparent and probably is tuned or not ? There is nothing to "tune" in the sense of a musical instrument .

Eastern designers often have a philosophy that allows a co-exisitence of being fully technical and objective down to minute details and being subjectivist "tinkerers" at the same time, without going split-personality.

 

[StefaanE]

Eastern designers often have a philosophy that allows a co-exisitence of being fully technical and objective down to minute details and being subjectivist "tinkerers" at the same time, without going split-personality.

And here I was thinking the guy was rather adept at using marketing-speech; know thine customer-base and suchlike.

 

[Mnyb]

Eastern designers often have a philosophy that allows a co-exisitence of being fully technical and objective down to minute details and being subjectivist "tinkerers" at the same time, without going split-personality.

It will be great fun for them to even try to verify their results or wait even to define the question ?

 

[Blumlein 88]

Eastern designers often have a philosophy that allows a co-exisitence of being fully technical and objective down to minute details and being subjectivist "tinkerers" at the same time, without going split-personality.

Examples???

 

[Phorize]

Emotion. Can an instrument measure the emotion of a performance or the emotional impact of a particular song or passage has upon an individual?

Obviously not, but why would you want it to?

 

[threni]

I must admit that I'm struggling to understand how it's possible to measure the audio via mics, computers, tape/disk etc during the performance and production, capturing it perfectly, but that somehow this same equipment is suddenly insufficient to measure the audio upon playback.