High Resolution Audio: Does It Matter?

[TSB]

Lost in that is one tester who managed to get 8 out of 10 right meaning there was 94.5% probability that he was identifying the proper source and not guessing. This is so close to 95% threshold that it should have been noted as significant and countering the larger conclusion but was not. Two other testers managed 7 out of 10 correct selections. These were all dismissed as exceptions and the total number of trials/listeners incorrectly relied upon.

This sounds like a misunderstanding of statistics.

If make 100 sets of 10 dice-throws I expect some of the sets to have 8/10 heads or 8/10 tails, even with a fair dice.

 

[kongwee]

You are confused.
This is a commonly quoted mistake.
There is no need for the waveform to be any more precise than is necessary to decode audible frequencies.

The audible parts of a 500Hz square wave are perfectly well reproduced with components at 1.5 kHz, 3kHz, 6kHz and 12kHz defining the shape of the wave and it looks square on an oscilloscope.
The audible parts of a 10kHz square wave are basically just a 10kHz sine wave, all higher frequency components of the square wave are inaudible so a 10hHz square wave sounds EXACTLY the same as a 10kHz sine wave, so the fact a 10kHz square waveform isn't kept at 44.1kH sampling frequency is not only correct technically but also for audibility.

You don't listen to and can't hear waveform.

Wait, I can heard square wave and sine wave perfectly, that is sound design. They sound different. There is tons of free software synthesiser that you can heard different waveform.

In oscilloscope you can see output, not a perfect square wave. flat mean zero frequency, rising and falling mean infinite frequency. Of course, it is no infinite frequency in real world. There is slew rate to judge out quick is the rise and fall, on to off or off to on position. Those with 500kHz amp oscillation draw better square wave than 350kHz in real world. You can see the vast different between tube and solid state doing the same square wave.

 

[danadam]

The audible parts of a 500Hz square wave are perfectly well reproduced with components at 1.5 kHz, 3kHz, 6kHz and 12kHz

To be precise, they are at 1.5k, 2.5k, 3.5k, ..., 19.5k. The components of square wave are f, 3f, 5f, ..., (2k-1)f.
Maybe a better example would be a 2k square wave, with components at 6k, 10k, 14k and 18k, which should sound the same as a square wave with additional 22k, 26k, etc. components, even though they look different:

 

[voodooless]

Wait, I can heard square wave and sine wave perfectly, that is sound design. They sound different. There is tons of free software synthesiser that you can heard different waveform.

Here is a 10K square and sine wave file, -2 dBfs, 24 bit 192 kHz. Have fun ABX'ing.

Edit: see a few posts down: seem one cannot trust Audition to create clean signals.

 

[kongwee]

To be precise, they are at 1.5k, 2.5k, 3.5k, ..., 19.5k. The components of square wave are f, 3f, 5f, ..., (2k-1)f.
Maybe a better example would be a 2k square wave, with components at 6k, 10k, 14k and 18k, which should sound the same as a square wave with additional 22k, 26k, etc. components, even though they look different:
View attachment 182282

16hz to 8kHz are fall into the music fundamental frequency. I really know square wave is made up of modulated sine wave. It is in textbook. Even in your graph, the peak and bumps can be heard in sound design in music frequency range. Why don't you give volts time domain of real world amplification.

I can almost make the waveform from your graph.

 

[danadam]

Here is a 10K square and sine wave file, -2 dBfs, 24 bit 192 kHz. Have fun ABX'ing.

That's not a band-limited square, so it is full of aliasing:

(and they are 16 bit, not 24)

 

[Frank Dernie]

To be precise, they are at 1.5k, 2.5k, 3.5k, ..., 19.5k. The components of square wave are f, 3f, 5f, ..., (2k-1)f.
Maybe a better example would be a 2k square wave, with components at 6k, 10k, 14k and 18k, which should sound the same as a square wave with additional 22k, 26k, etc. components, even though they look different:
View attachment 182282

My poor sums. Too early in the morning.

 

[Frank Dernie]

Wait, I can heard square wave and sine wave perfectly, that is sound design. They sound different. There is tons of free software synthesiser that you can heard different waveform.

In oscilloscope you can see output, not a perfect square wave. flat mean zero frequency, rising and falling mean infinite frequency. Of course, it is no infinite frequency in real world. There is slew rate to judge out quick is the rise and fall, on to off or off to on position. Those with 500kHz amp oscillation draw better square wave than 350kHz in real world. You can see the vast different between tube and solid state doing the same square wave.

It depends on frequency, a 500Hz square wave sounds quite different to a 500Hz sine wave because a lot of its harmonic components are at audible frequencies.
A 10kHz square wave sounds exactly the same as as a 10kHz sine wave because none of its harmonic components are at audible frequencies.

The difference between valve and SS amps on low frequency square waves can be extreme due to phase shift but IME I can't hear phase, so when I did the experiment on myself about 50 years ago I was astonished to watch huge change in wave shape when I changed phase but heard no difference at all. I was so surprised I repeated the evaluation a couple of times and got others to experience it too.
Loudspeakers almost all created appalling square wave shape change, thankfully people don't notice IME

 

[kongwee]

It depends on frequency, a 500Hz square wave sounds quite different to a 500Hz sine wave because a lot of its harmonic components are at audible frequencies.
A 10kHz square wave sounds exactly the same as as a 10kHz sine wave because none of its harmonic components are at audible frequencies.

For my signal generator in DAW, both signal is almost the same only with anti aliasing on square wave. That is in digital domain. Another big topic on anti aliasing which not my expertise.

 

[danadam]

That's not a band-limited square, so it is full of aliasing

Here's something valid: https://mega.nz/folder/R5hVAYZZ#kXYcNPNOnXMbCBzBvnMCCA

 

[voodooless]

That's not a band-limited square, so it is full of aliasing:
View attachment 182308 View attachment 182309

(and they are 16 bit, not 24)

That’s very strange, blame Adobe Audition it even looked correct in the image preview of the generator.. did not check the result afterward though. It seems that you can’t blindly trust these software tools..

 

[Frank Dernie]

For my signal generator in DAW, both signal is almost the same only with anti aliasing on square wave. That is in digital domain. Another big topic on anti aliasing which not my expertise.

I am pointing out whatever it looks like, which is irrelevant since we don't listen with our eyes, the SOUND of a 10kHz square wave is the same as a 10kHz sine wave. Unless you can hear 30kHz of course.
There is absolutely no point in either trying to do a sound recording of something nobody could hear or to reproduce it.

That is like concluding that a photographic lens is inadequate because it can't focus X-rays in the same plane as visible light.

 

[radix]

Don't most DACs have filters that would make 30 kHz playback impossible anyway? For example, the Topping filters (dx3pro+) are all down at least -10dB by 22 kHz. And even the most generous is -80 dB at 28 kHz.

https://www.audiosciencereview.com/forum/index.php?attachments/topping-dx3-pro-measurements-filter-usb-dac-bluetooth-stereo-png.157581/

 

[voodooless]

Don't most DACs have filters that would make 30 kHz playback impossible anyway? For example, the Topping filters (dx3pro+) are all down at least -10dB by 22 kHz. And even the most generous is -80 dB at 28 kHz.

https://www.audiosciencereview.com/forum/index.php?attachments/topping-dx3-pro-measurements-filter-usb-dac-bluetooth-stereo-png.157581/

That's at 44.1 kHz sampling only. These filters are dependent on the sample rate.

 

[earlevel]

I used to listen to vinyl, despite all the hiss and cracks. Resolution beat any CD. The soundstaging and layering are unmatched. 24 bit 196kHz was to beat vinyl in this expect, even DSD. Put an orchestra recording, you will heard the smearing effect of Red Book. Vinyl has relevant 20bit 170kHz in digital expect. Of course this generation of audio, every hard to find new AAA recording that explore the advantage of vinyl. DVD disc that carry these resolution just dead.

Dithering or noise shaping is used when pros are working higher than 16bit 48kHz and downsample to suit the Red Book.

I have a theory—well, I at least wonder if these kind of glowing descriptions of vinyl ("the soundstaging and layering are unmatched", etc.) is precisely because of "despite all the hiss and cracks". I think we have to let our brains ignore the crackles and hiss and imagine the purity hidden underneath. And it sounds so good in our minds. 24-bit digital audio hides nothing, we're left to hear the actual recording, in all its glory and flaws, no excuses. Kind of like when I got my first HD tv and realized the nice news anchor lady was a lot older and wrinkled looking than it had seemed before (lol).

Similarly, in the recording industry, Lexicon hardware reverbs where highly revered (and still prized to this day, though they have much more able competition). These reverbs had to get by on limited arithmetic capabilities, word sizes that people cringe at today. As a result, the reverberation they produced lost resolution quickly. But apparently, to a lot of people, it sounded like it faded to infinity with great clarity...as it in-fact faded into noise. Yet many people prized the incredible apparent resolution and "depth" of this devices, imagining the reverb decaying to infinity, even though it's clear they suffered in this respect.

On your final comment, about dither, it's really not about sample rate conversion, which is typically done with higher precision that 24-bit. Dither is all about reducing that result to 24-bit or 16-bit (any bit reduction, though at 24-bit it's not hearable, but what the heck).

 

[kongwee]

On your final comment, about dither, it's really not about sample rate conversion, which is typically done with higher precision that 24-bit. Dither is all about reducing that result to 24-bit or 16-bit (any bit reduction, though at 24-bit it's not hearable, but what the heck).

I just do for redbook anyway. I doesn't cost me an arm and a leg to down sample from my DAW to suit CD.

 

[Frank Dernie]

is precisely because of "despite all the hiss and cracks"

I have mentioned this before but when we id an investigation 20 odd years ago into why LPs didn't sound too bad when considering how poor the technical performance is compared to CD two things stood out to me.
Adding extra noise to a digital file gave the impression of a bigger sound stage.
Reducing crosstalk to 35dB didn't make any difference.

 

[Deleted member 16543]

I have mentioned this before but when we id an investigation 20 odd years ago into why LPs didn't sound too bad when considering how poor the technical performance is compared to CD two things stood out to me.
Adding extra noise to a digital file gave the impression of a bigger sound stage.
Reducing crosstalk to 35dB didn't make any difference.

I think the main benefit of LPs actually resides in forcing the master to not being ridiculously loud, which is something easily achievable instead (and to me inexplicably sought after) with digital.
The relation of added noise and increased sound stage is very interesting, though. I wonder if that's the case with binaural recordings too.
Did increasing crosstalk higher than -35 dB pull the image back to the center?

 

[Frank Dernie]

I think the main benefit of LPs actually resides in forcing the master to not being ridiculously loud, which is something easily achievable instead (and to me inexplicably sought after) with digital.
The relation of added noise and increased sound stage is very interesting, though. I wonder if that's the case with binaural recordings too.
Did increasing crosstalk higher than -35 dB pull the image back to the center?

I think the compression problem is more recent, probably no more than the last 20 years.
Dynamic music needed to be compressed to an extent for LP to fit into the limited dynamic range of the medium.
This sin't necessary with CD and in the early days we had some great recordings.

Personally I don't know when the "loudness wars" actually started but pop music recordings with the dynamics considerably compressed so the average loudness can be high without clipping is everywhere. I suspect that it always sounds more dynamic if compared to an uncompressed original without adjusting the volume control to get the average loudness the same because louder always does. It actually wouldn't be hard to cut these restricted dynamics recordings onto an LP but thankfully they usually don't.

I was just a participant not running the test and I don't remember anything other than CD separation compared to typical excellent cartridge separation being compared.
The noise test didn't add enough to be normally audible nor obvious in very quiet passages.

 

[voodooless]

Personally I don't know when the "loudness wars" actually started but pop music recordings with the dynamics considerably compressed so the average loudness can be high without clipping is everywhere.

Look here for some history and analysis.