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

[radix]

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 used 96 kHz on a Focusrite Scarlett 2i2 in the experiment.

 

[pma]

They used 96 kHz on a Focusrite Scarlett 2i2 in the experiment.

Yes I know.

 

[lowgain]

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..

I did some blind testing with the best gear I am willing to spend money on (Topping dx3pro+, HD660s), in the quietest environment I could find in the house, and can't tell the difference between any reasonable bitrate MP3 or AAC and lossless.

I then proceeded to test Opus, and I can't hear any difference between 96kbps and lossless. Even at 64kbps I had difficulty hearing a difference.

I also played around with the 2L hi res test files and downsampled to 8bits 24kHz with dithering, noisy but sounded still good. (I didn't ABX any of these, just curious to hear what it sounds like if you really start to look for bad quality).

Needless to say, stating this on most fora will get you laughed at.

Edit: a while ago I compared Spotify to tidal lossy, see what you get for a similar price:

 

[Talisman]

I did some blind testing with the best gear I am willing to spend money on (Topping dx3pro+, HD660s), in the quietest environment I could find in the house, and can't tell the difference between any reasonable bitrate MP3 or AAC and lossless.

I then proceeded to test Opus, and I can't hear any difference between 96kbps and lossless. Even at 64kbps I had difficulty hearing a difference.

I also played around with the 2L hi res test files and downsampled to 8bits 24kHz with dithering, noisy but sounded still good. (I didn't ABX any of these, just curious to hear what it sounds like if you really start to look for bad quality).

Needless to say, stating this on most fora will get you laughed at.

Rest assured, this is luckily not the kind of forum where they will tell you it's just because your sound system sucks and with their thousands of euro systems you could easily hear the differences day and night.
I also suspect that stripped, blind, in the face of decent lossy and lossless files, they would have the same difficulty as you and I.
But if they can look they can easily feel the difference of the stones on the CD transport and can easily feel the veil being lifted.

 

[lowgain]

I once hung a veil over my speakers and could not tell the difference between lifted and not lifted.

Since then, whenever I here someone say it sounds as if a veil had been lifted, I think they mean it sounds the same.

 

[antcollinet]

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.

Also : Digital volume control.

 

[j_j]

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.

Yeah, basic physics is a serious impediment to getting beyond 20 bits. Remember shot noise, for instance, and figure out how many electrons/second you need to reach each bit, and convert that into milliamps. The results are illuminating to say the least.

 

[j_j]

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



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.

Fine structure does go away, but ENVELOPE sensitivity does not go away. Many people forget this. It is hard to localize a 10kHz sine-wave signal. It is easy to localize a pulsed 10kHz signal, either with a gaussian pulse, a minimum-phase pulse, or a chopper. Easily shown. Using the gaussian pulse can assure you that you are avoiding any substantial out of band signals (i.e. using one with 2-sigma pulse width (in frequency) of 4000Hz ensures you're down 5 sigma (and that's a lot!!!) at 20khz.

 

[j_j]

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

Please be careful to ensure that your various "transients" are actually bandlimited properly, and that any digital reconstruction is properly anti-imaged.

When you do this, you will find promptly that there is much more resolution than 1/fs, in fact, you'll find out 'fs' is not the issue, the issue is total system bandwidth and SNR, which for any given point on the cochlea is under 5khz and 30dB. You can't avoid finding out, then, that there will be clear information remaining in the 44.1 kHz sampling system that demonstrates very sub-sample-interval time resolution.

It's there. If you don't find it, you made a mistake.

 

[Deleted member 48726]

I did some blind testing with the best gear I am willing to spend money on (Topping dx3pro+, HD660s), in the quietest environment I could find in the house, and can't tell the difference between any reasonable bitrate MP3 or AAC and lossless.

I then proceeded to test Opus, and I can't hear any difference between 96kbps and lossless. Even at 64kbps I had difficulty hearing a difference.

I also played around with the 2L hi res test files and downsampled to 8bits 24kHz with dithering, noisy but sounded still good. (I didn't ABX any of these, just curious to hear what it sounds like if you really start to look for bad quality).

Needless to say, stating this on most fora will get you laughed at.

Edit: a while ago I compared Spotify to tidal lossy, see what you get for a similar price:

Do you care to explain what yo look for?
It looks like CD and spotify look more alike than Tidal where it looks like information is missing in high frequencies. Is this Tidal HiFi or High setting? Spotify on Very High?

 

[lowgain]

Do you care to explain what yo look for?
It looks like CD and spotify look more alike than Tidal where it looks like information is missing in high frequencies. Is this Tidal HiFi or High setting? Spotify on Very High?

Don't remember details, but everything is in my original investigation here: https://numericalnoob.blogspot.com/2019/06/streaming-services-have-become-my-main.html

Some side notes:
- I was mainly concerned back then that a service like Tidal with a pricing structure around quality could degrade the lower tier to the extent that a significant portion would buy the higher quality tier, as that makes economic sense.
- When Apple made everything the same price, this became less of a concern

 

[Tangband]

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.

Yes, and its good to understand ( as you know ) that a 24 bit digital resolution IS 24 bit , not less - in the digital domain and when working with a DAW. Its the analog stages that attenuate the SINAD, and you have to have those to go from analog to digital or the reverse.

A processor with 32 bit ability is 32 bit in digital domain , not less. The same is true regarding true 24 bit digital volume controls.

As you know , when recording real instruments, the microphones themselves have a noise floor no better than - 100 dB , at best. Often less, about -80… worse than any modern A/D.

 

[Tangband]

This is a measurements of the noise level in a real concert hall recorded with a 24 bit A/D using microphones.
As one can see, the room is never 100 % quiet and the microphones has worse noise level at the low frequencies.

 

[pma]

This is a measurements of the noise level in a real concert hall recorded with a 24 bit A/D using microphones.
As one can see, the room is never 100 % quiet and the microphones has worse noise level at the low frequencies.

No SPL calibrated Y axis? What kind of microphone? Large or small diaphragm?

 

[j_j]

This is a measurements of the noise level in a real concert hall recorded with a 24 bit A/D using microphones.
As one can see, the room is never 100 % quiet and the microphones has worse noise level at the low frequencies.
View attachment 261462

I'd be looking for that 370/380Hz noise, too. That looks pretty coherent, like machinery or something.

 

[Tangband]

No SPL calibrated Y axis? What kind of microphone? Large or small diaphragm?

faktiskt.io • Visa tråd - 24-192 bättre än 16-44?

 

[Deleted member 48726]

Don't remember details, but everything is in my original investigation here: https://numericalnoob.blogspot.com/2019/06/streaming-services-have-become-my-main.html

Some side notes:
- I was mainly concerned back then that a service like Tidal with a pricing structure around quality could degrade the lower tier to the extent that a significant portion would buy the higher quality tier, as that makes economic sense.
- When Apple made everything the same price, this became less of a concern

So it was tested Spotify at its highest setting and Tidal at it's highest lossy setting "high" in the desktop application.
CD // Spotify "Very high" // Tidal "High"

Maybe you should rename your uploaded pictures so there can be no misunderstandings what is compared.

 

[markanini]

Comparing lossy audio by graphs is flawed. It says nothing about detectable differences in a DBT with a PCM reference.

 

[j_j]

Comparing lossy audio by graphs is flawed. It says nothing about detectable differences in a DBT with a PCM reference.

To some extent I disagree, but one has to be someone who built a perceptual codec to do much. But yes, testing a perceptual codec requires perception.

 

[j_j]

This is a measurements of the noise level in a real concert hall recorded with a 24 bit A/D using microphones.
As one can see, the room is never 100 % quiet and the microphones has worse noise level at the low frequencies.
View attachment 261462

Perhaps it's time to remind people (not necessarily you) that the actual noise due to air molecules hitting your ear drum (which is what makes "pressure" exist, of course) is somewhere in the neighborhood of 6 to 8 dB SPL (flat weighted). No, you can't hear that, barely, because of the white spectrum of that particular noise (the noise floor here has some other components, which is common everywhere, it's hard to block low frequencies) has little energy in any given ERB.

But for 20-20K that's a solid minimum noise level. The only way to get rid of that noise is to remove the air from both sides of the eardrum, and I suspect the listener would be upset, at the very least. It's not an experiment to actually attempt.