Is high-resolution audio audible or not audible and a waste of data?

[pma]

No, now that is NOT quite an accurate statement.

I admit I cannot get into expert arguing on this subject, what I meant is described e.g. here:

Smallest perceivable interaural time differences

It is well-established that the smallest discrimination thresholds for interaural time differences (ITDs) are near 10 μs for normal hearing listeners. However,

doi.org

 

[j_j]

I admit I cannot get into expert arguing on this subject, what I meant is described e.g. here:

Smallest perceivable interaural time differences

It is well-established that the smallest discrimination thresholds for interaural time differences (ITDs) are near 10 μs for normal hearing listeners. However,

doi.org

They need to use a broadband pulse with a very sharp attack. (Well, that's redundant, really, but above 2 kHz there's not much detection of the sine wave phase, but there is substantial detection of pulse arrival, including the "sharpness" of the pulse in one ear.

While it is important to have subjects with decent hearing, I must also say that training and learning what to listen to is very important. This is why I have objected to any number of tests that were just thrown at subjects.

I must also note that they did not include frequencies above the usual range for ITD's, which was a mistake, assuming they used signals with a repeated short onset. Above 2kHz I agree there is not much (any?) response to interaural phase of a sine wave, or a series of tones, but now try a gaussian pulse centered at, say, 6kHz with a +-4 kHz bandwidth, or even better a minimum-phase pulse with similar bandwidth. This kind of signal will expose ITD sensitivity that is repeatedly denied, the key being that at higher frequencies, it's not the phase of a tone, but the onset of a signal envelope. This is also easy to test in matlab.

 

[MattHooper]

"We don't need no steenking hi-res!"

 

[j_j]

Another thing is that the transients whose rise rate is lower than 22usec are diluted in time, before and after the event, as reported in the graph of the first post. And since our ear has a higher sensitivity than this value (6-10usec), 44KHz is not physically sufficient to preserve transient events for our ears.

Any transient that occupies less than two samples in the MAIN LOBE (not the entirety) at a sampling rate (any rate) is not properly anti-aliased. This is neither an option nor a guess, its simple mathematics.

 

[j_j]

I just don't see how you can equate sensitivity to ITD with sensitivity to rise time.

Well, the way you do it is to apply the noise performance of the ear to the bandwidth of the pulse (fast rise time == wider band). Oddly, this predicts what's actually observed.

The old CRTs were around 15kHz (15.75kHz? I think) horizontal sweep. Some modern ones are at about 31kHz, but I very much doubt you hear that

B&W 15750, NTSC 15735.

 

[pma]

With a simple calculation it can be seen that using a sampling frequency of 44KHz, we will be able to correctly represent transients no shorter than 22μs; If they have a shorter duration, these will be "spread" over time. This aspect can be seen in the following figure.

Time difference between channels is maintained even if transient is faster than the rise time corresponding to the sampling frequency.

44.1 kHz sampling ADC, red channel delayed by 5us approx. by 1k+5nF RC, simultaneous sampling

192 kHz sampling ADC, red channel delayed by 5us approx. by 1k+5nF RC, simultaneous sampling

44.1 kHz sampling ADC, 5kHz sine, red channel delayed by 5us approx. by 1k+5nF RC, simultaneous sampling

192 kHz sampling ADC, 50kHz sine, red channel delayed by 5us approx. by 1k+5nF RC, simultaneous sampling

5us/div time axis. Please note that RC time delay is getting shorter at higher frequencies (see GD as well)

Note: ADC captured files are then played through DAC and DAC analog output is digitized at 25MS/s (25MHz) sampling rate.

50kHz sine spectrum B 5us delay 192kHz ADC sampling.png, 25MHz analog output sampling

 

[pma]

"We don't need no steenking hi-res!"

There are situations when 24-bit ADC resolution is good to have.

 

[mhardy6647]

There are situations when 24-bit ADC resolution is good to have.

Say more.

 

[fieldcar]

This appears to be spreading the whopper that the time resolution of a PCM system is limited to its sampling period again. Please, enough of that. It's just not a valid comparison.

I think people keep failing to realize what you mean despite repeating yourself. You even did a nice matlab experiment on another thread showing slight timing differences shifting the lobes/sampling intersections up and down to correspond with phase shift. Timing resolution does not have direct correlation with sampling frequency alone. Timing resolution is a result of both bit depth and sampling frequency.

Good visual from the Monty video illustrates this perfectly. Redbook audio has far greater than 22us time resolution.

https://imgur.com/p4pdcj9

 

[Deleted member 48726]

I don't know if anyone is capable of feeling such differences. I can't distinguish a 320 mp3 file from FLAC, imagine if I can distinguish 16/44 from 24/96.
Music in 16/44 is all I need, but only for the psychological tranquility of having lossless files, but the reality is that a 320kbs mp3 already has all the audio quality that I can perceive

Oh yes. Same here. I'm back and forth whether to switch to Tidal from Spotify or not. I know for a fact that I can't tell 320 kbps OGG Vorbis from lossless with headphones. But as you I may not find inner peace without..

 

[kemmler3D]

Say more.

If you're recording audio it's nice to not worry about headroom due to quantization noise, anyone recording in 16 vs. 24 these days probably has a screw loose. Now, whether the ADC actually delivers 24 whole bits of SNR, doubtful, but I'll take as many as I can get.

 

[fpitas]

Heck, even my DX-2006 has a 40 bit DSP for the math heavy lifting.

 

[mhardy6647]

My belief/understanding is that 24-bit makes good sense for the original recording, mixing, and mastering processes.
I can rationalize using higher resolution upstream of the product -- just as I can rationalize the use of larger format cameras and finer-grain film to take photographs that would ultimately be (mass) reproduced for public consumption in magazines or what-have-you.

 

[Blumlein 88]

If you're recording audio it's nice to not worry about headroom due to quantization noise, anyone recording in 16 vs. 24 these days probably has a screw loose. Now, whether the ADC actually delivers 24 whole bits of SNR, doubtful, but I'll take as many as I can get.

Pretty easy to get 17 bits with quite a few doing 18 bits of SNR in an ADC. A few do better, but not many get better than 19 bits, and I'm not sure I know of any regular audio ADCs that get more than 20 bits.

 

[markanini]

My belief/understanding is that 24-bit makes good sense for the original recording, mixing, and mastering processes.
I can rationalize using higher resolution upstream of the product -- just as I can rationalize the use of larger format cameras and finer-grain film to take photographs that would ultimately be (mass) reproduced for public consumption in magazines or what-have-you.

If audiophools were photo enthusiasts, they would unironically believe their hawk eyes made it possible to see beyond the resolution of the magazine print.

 

[mhardy6647]

If audiophools were photo enthusiasts, they would unironically believe their hawk eyes made it possible to see beyond the resolution of the magazine print.

... and then there's the great practical joke of optics: empty magnfication.

d = λ/(2NA)

Microscope Resolution: Concepts, Factors and Calculation

This article explains in simple terms microscope resolution concepts, like the Airy disc, Abbe diffraction limit, Rayleigh criterion, and full width half max (FWHM). It also discusses the history.

www.leica-microsystems.com

ZEISS Microscopy Online Campus | Microscopy Basics | Numerical Aperture and Resolution

The numerical aperture of a microscope objective is the measure of its ability to gather light and to resolve fine specimen detail while working at a fixed object (or specimen) distance.

zeiss-campus.magnet.fsu.edu

... and that's why we have electron microscopes

 

[fpitas]

... and then there's the great practical joke of optics: empty magnfication.

d = λ/(2NA)

Microscope Resolution: Concepts, Factors and Calculation

This article explains in simple terms microscope resolution concepts, like the Airy disc, Abbe diffraction limit, Rayleigh criterion, and full width half max (FWHM). It also discusses the history.

www.leica-microsystems.com

ZEISS Microscopy Online Campus | Microscopy Basics | Numerical Aperture and Resolution

The numerical aperture of a microscope objective is the measure of its ability to gather light and to resolve fine specimen detail while working at a fixed object (or specimen) distance.

zeiss-campus.magnet.fsu.edu

... and that's why we have electron microscopes

Just don't get physicists talking about the cables!

 

[Pinox67]

Hmm, can you elaborate or cite something on this point? I don't totally understand. How would a steeper transient arriving at the same time carry spatial information that the lower-bandwidth transient doesn't, if we can't actually perceive the extra frequency content?

The filtering of the ear before the signal reaches the brain would look something like the digital filtering in your first post, right?

Also: consider that if timing information were limited to 5 microseconds resolution in a 44.1khz file, the pitch of sounds would be quantized and aliased terribly. You could only represent pitches that were a multiple of 5 microseconds. This is, as we know, not something that happens in digital audio.

I'm preparing simulations to answer these questions. It will take a bit 'of time...

 

[MattHooper]

I've never chased listening to hi-res audio. Perhaps it's idiosyncratic, but for me I want to optimize my system for the sources that I will be most regularly listening to, which is going to be regular old CD quality and vinyl. I don't think I'd hear much difference with most hi-res audio (that is only by virtue of their being higher resolution than CD), and to the extent I did, I wouldn't care that much as it doesn't really make a dent in my general musical diet.

 

[kemmler3D]

Relevant: https://www.frontiersin.org/articles/10.3389/fnins.2014.00034/full

Because ITD information becomes increasingly misleading as the frequencies and azimuths increase, there would be survival value in a binaural system that becomes insensitive to ITD at moderately high frequency. Such a system would defend its owner from dangerous localization cues that could lead to mislocalization. In fact, there is unequivocal evidence that fine-structure ITD sensitivity disappears at about 1500 Hz. The upper limit of ITD sensitivity was explored by Zwislocki and Feldman (1956) and by Klumpp and Eady (1956), who found an upper limit of 1300 Hz. Mills (1958) found a limit of 1400 Hz, and Nordmark (1976) found 1430 Hz.

I found another paper that looked at ITD performance vs. frequency bandwidth of impulses... https://link.springer.com/article/10.1007/s10162-011-0303-2 Interestingly it went up beyond 1500hz so I'm not sure what to make of this, I may have misunderstood something.