What objectivists and subjectivists can learn from each other

[rbbert]

There's a big difference between being able to measure everything and actually doing it (remember TIM) and another big difference between having the measurements and knowing what they mean in audio terms.

As I posted back near the beginning of this thread (or maybe it was its predecessor), the whole idea that we are very close to being able to reproduce live music with our playback systems is objectively by any measure far from the truth, and it's NOT the difference between stereo and multi-channel. If our hardware were as good as some here seem to think, there would be moderately priced multi-channel audio systems that could play a well made recording as nearly indistinguishable from the original music, and that's just not believable.

 

[Ethan Winer]

Dynamic Range:
48Khz: 153 dbA
96Khz: 150 dbA
192Khz: 148 dbA
384Khz: 145 dbA

Yes, but that's with A-weighting which intentionally skews the numbers to more closely match how we hear. Using A-weighting for assessing audibility is not unfair, but it's not what I was addressing which is pure thermal noise over the audible band.

--Ethan

 

[Ethan Winer]

There's a big difference between being able to measure everything and actually doing it (remember TIM) and another big difference between having the measurements and knowing what they mean in audio terms.

Well, I certainly know what the measurements mean! As for TIM, that would be revealed easily in a null test. But you are correct that past measurements were not always complete. Heck, they're still not complete as usually published, even though it would be trivial to do that. BTW, there's nothing magical or elusive about TIM. It could have been shown long ago by simply measuring the distortion of a 20 KHz sine wave burst at full power.

the whole idea that we are very close to being able to reproduce live music with our playback systems is objectively by any measure far from the truth

Again, this has nothing to do with the fidelity of the gear, and I wish people would stop inserting this red herring into a discussion of audible transparency and differences between audio devices. It's another issue entirely, having more to do with loudspeakers, room acoustics, and mic technique than audio fidelity.

--Ethan

 

[Bruce B]

Yes, but that's with A-weighting which intentionally skews the numbers to more closely match how we hear. Using A-weighting for assessing audibility is not unfair, but it's not what I was addressing which is pure thermal noise over the audible band.

--Ethan

So Ethan.... what is the "thermal noise over the audible band" of the MSB Studio ADC? Also... what is the "non" A-weighted measurements of the MSB Studio ADC. If you don't know.... don't respond. Besides, the graph spectrum shows the noise from 7.5k to 14.5k.. certainly within the "audible band".

 

[rbbert]

... Again, this has nothing to do with the fidelity of the gear, and I wish people would stop inserting this red herring into a discussion of audible transparency and differences between audio devices. It's another issue entirely, having more to do with loudspeakers, room acoustics, and mic technique than audio fidelity.

--Ethan

Statements like this make it hard to take your posts seriously, much as I would like to

 

[Ron Party]

Statements like this make it hard to take your posts seriously, much as I would like to

Why? Is it up for discussion the distinction to which Ethan refers but which may be blurred?

 

[andy_c]

Ethan, that should be "quantization noise", not "thermal noise". In a proper A/D, that would be dithered quantization noise.

Also, it's a common error to take an FFT plot, take a full-scale signal and its difference from the "hash" level of the plot, and call this the signal-to-noise ratio. It's not. With the FFT, the level of the "hash" depends on the number of FFT points one uses. Picking an FFT size determines the number of frequency bins. The larger the number of FFT points, the narrower the FFT frequency bins. For a given signal and RMS noise level, with white noise, one can get lower and lower "hash" levels in the FFT by simply increasing the number of points. This is because the frequency bins become narrower and narrower, inversely proportional to the number of FFT points, with a consequently lower per-bin noise level.

Here are two examples. The first is a full-scale 100 Hz sine wave, with triangular PDF dither of +/-1LSB added, quantized to 16 bits and sampled at 44100Hz. The FFT is taken over 100 cycles of the waveform.

Now here is the exact same waveform with FFT taken over 1000 cycles.

There's an apparent 10dB noise decrease. But it's the same signal, with the same noise! This shows the dependence of FFT "hash" noise level on the number of FFT points.

These plots were done with GNU Octave, a freeware Matlab workalike.

Edit: I should add that there's no averaging or windowing done in this plot. FFT leakage is prevented by always taking the samples over an integer number of cycles (minus one sample at the end). Each cycle also has the same number of samples (=441).

 

[rbbert]

I've never met anyone who isn't tone-deaf (and even some who are) who can't quickly tell the difference between live acoustic music and recorded music, no matter what the reproduction system or setting. (Imperfect) Transducers undoubtedly play a major role in this setting, but it is highly likely (almost a certainty) that the recording medium (with its associated imperfections) also plays a significant role. To simply state that it doesn't diminishes the poster's credibility significantly.

 

[Ron Party]

I think you've misunderstood that which he posted.

 

[rbbert]

I think you've misunderstood that which he posted.

Why do you believe that no current converters are audibly transparent? On what do you base that? Your own proper blind tests? Are you aware that a major pro audio converter manufacturer did a test where they re-recorded a piece of music ten times in a row, and even after ten generations nobody was able to tell the original from the output? I forget the vendor, and this was a few years ago. But that alone proves to me that one A/D/A pass can be truly transparent. So then the question is as I asked: Why do you believe otherwise?

--Ethan

What part of this didn't I understand?

 

[Ron Party]

I now see from where you are coming. I did not realize your post was made with respect to that passage.

 

[rbbert]

I now see from where you are coming. I did not realize your post was made with respect to that passage.

It seems to me that this is what this whole topic is about. There is a fundamental logical flaw in the position Tim and Ethan have taken.

 

[Ron Party]

Now you lost me.

 

[NorthStar]

I am reading so many threads in these forums here,
and I am getting acquainted more and more with the members.

For me, discussions are a mean to learn, to advance in my natural knowledge's progression.

I respect EVERYONE here, without a single exception.
And I love challenges, and lively discussions that are from several opinion's sides.
Because that's how we learn and interact between all of us as true and mature adults.

It amazes me that the Internet can be so emotional sometimes, but then human reality stands ...

This Audio business of ours (our livelyhood, our deep passion...), our love of the Music,
our passionate appreciation for the artists that gave us the music we cherish,
and the motion pictures that create deep emotions;
all of this is pure creation, emotions sharing, interpretations too, and so much more ...

And why, oh why, sometimes do we have to stumble?
Because very simply we are just humans, an animal race with imperfected aptitudes.
We are in constant evolution, and we are having the toughest time of our lives
to improve, to advance, to learn from our mistakes, to love more than hate.

Audio is one of the most debated subject because of its various temperamental rainfalls.
It has so many unexplained complexities that the sides we are on are just a glimpse of the entire true picture.

I just had to say what I just said, so please bear with me. Thank you.

 

[rbbert]

Now you lost me.

And I have not heard of an amplifier that does not have a noise floor. That doesn't mean I can hear it. And while I agree that there can be a correlation between measured impulse response and sound, in competent, much less professional or high-end gear, it falls into the realm somewhere between insignificant artifacts we have to train ourselves to hear, and unidentifiable with your eyes closed. I believe in jitter. I even believe in audible jitter. But much more importantly, I believe the audible differences between DACs, 99.9% of the time, are in their analog output stages. There is a whole world full of gearheads out there, chasing the boogeyman of inaudible jitter from under their digital beds while warmly embracing analog output "voiced" right out of fidelity to the recording. But even that's rare. What's common is expectation bias.

Tim

We all listen to "analog", obviously, that's why we have ADCs and DACs if the storage medium is digital. Now, does Tim mean that despite the earlier (consistent) claim that a well-designed and well-built (even if relatively inexpensive) amplifier is "perfect" (or audibly so, at least), that now differences in DACs are due to those types of amplifier stages? And what kind of "voicing" are we talking about; something trying to mimic whatever imperfections of vinyl LPs one likes? Or similarly of analog tape (which is quite a different animal from vinyl LPs, of course)? And has he (or anyone) measured these differences in what are essentially SS amplifiers (i.e., the analog output stage of a DAC) and correlated those measurements with the different "sound" of different DACs?

That's what I mean by a fundamental logical flaw.

 

[Bruce B]

I believe the audible differences between DACs, 99.9% of the time, are in their analog output stages. Tim

I wouldn't give it that much. While the analog output stages do contribute alot to the overall sound of an A/D or D/A, I'd also put emphasis on the clock and how and the implementation of the interface.

 

[fas42]

As I posted back near the beginning of this thread (or maybe it was its predecessor), the whole idea that we are very close to being able to reproduce live music with our playback systems is objectively by any measure far from the truth, and it's NOT the difference between stereo and multi-channel. If our hardware were as good as some here seem to think, there would be moderately priced multi-channel audio systems that could play a well made recording as nearly indistinguishable from the original music, and that's just not believable.

The good news for everyone is that this is wrong. The bad news is that this is relatively difficult to do, not because moderately priced gear is not inherently capable of it, but because a great deal of care and attention to detail has to be applied. Exactly the sort of thing that manufacturers of consumer grade products have well in hand ... :b

One thing that is not really appreciated here is that the ear/brain mechanism does not work on a smooth path of capability: in other words, it is not a case that an audio system will just keep sounding a little bit less unreal as it improves in some, or a number of technical, measurable areas. To use a good, "scientific" term there is a discontinuity in the functioning of the hearing mechanism: below a certain standard it sounds unreal, just hifi; but above that standard, the hifi aspect evaporates, and it becomes difficult to "prove" to your body that it is not real.

Why this sort of thing is not shouted from the rooftops is that the level of quality required, in certain key areas, is poorly understand, and quite difficult to achieve: engineers and audio professionals just tossing together a heap of very expensive equipment are highly unlikely to fluke it -- because they don't appreciate the subtleties involved. Most times, people stumble across it by accident at odd times, and find it very hard to replicate.

Frank

 

[NorthStar]

'Accuracy', we've talked 'bout this before. What is it?

What is accurate reproduction?

For me it's the constant search, the unobtainable.

 

[fas42]

Audio is one of the most debated subject because of its various temperamental rainfalls.
It has so many unexplained complexities that the sides we are on are just a glimpse of the entire true picture.

I just had to say what I just said, so please bear with me. Thank you.

Right on, Bob ... :b

Frank

 

[NorthStar]

Music is emotions, and you simply cannot measure it.