Inside High res Music (Antonio Forcione) (Video)

[Thb2021]

If you've ever recorded a computer monitor with a video camera you've no doubt seen moving lines across the monitor in the video. That's because the "sampling rate" (shutter speed) of the camera is 29.97 per second, while the monitor might be refreshing at 60 or 120hz. That's a time difference. It's not because one is capturing a higher wavelength (frequency) of light particles than the other.

A timpani drum tops out at 200hz approx. By that same logic, a 400hz sample rate is all that is needed to perfectly capture a timpani. Throw in harmonics and let's say the full wave has meaningful energy up to 1K to be generous. Do you think a 2khz sample rate will sound identical to 44.1, with everything else being empty space or noise? How about 8khz?

Obligatory reading and viewing: https://www.audiosciencereview.com/...lution-of-redbook-16-44-pcm.22102/post-733281

That video is not the same thing as what I'm saying at all.

If resolution doesn't exist then a timpani sampled at 2khz should sound identical or better than one sampled at 44.1.

 

[voodooless]

What do you specifically want to know &/or understand?

Well.. I basically already put forth a bunch of questions, so let’s start from there

 

[voodooless]

If you've ever recorded a computer monitor with a video camera you've no doubt seen moving lines across the monitor in the video. That's because the "sampling rate" (shutter speed) of the camera is 29.97 per second, while the monitor might be refreshing at 60 or 120hz. That's a time difference. It's not because one is capturing a higher wavelength (frequency) of light particles than the other.

No, that’s aliasing. Besides, you cannot perfectly compare video sampling with audio sampling. In video every discrete time interval is a accumulation of the light on the pixel during that interval (or a fraction of it). In audio, the sample is only the value at the exact point in time. It does not say anything of what was before or after.

That video is not the same thing as what I'm saying at all.

Then what are you saying?

If resolution doesn't exist then a timpani sampled at 2khz should sound identical or better than one sampled at 44.1.

If all your assumptions are correct, then yes, it should sound the same. That is what the sampling theorem is all about. What of it? Sounds very much like an argument from incredulity.

 

[testp]

i find it amazing how everything above 24 Khz is left into the signal in the first place and that bigger bandwith for high-res is not used for listenable music, let's say 24 Khz.

The way i would have done it: multiply the space for every Khz, so 24 Khz gets x 2 space for 48 Khz, 96 Khz get's 24 Khz x 4 and so on...

so in this theory, every hertz would take double the space for storing it when recorded ( in 48 Khz -> 1 Hz = 2 Hz : decoder just needs to know that dividing is needed)
this way it would make sense that the bandwith of music is strecthed into bigger bandwith storage, but i guess that is not true in High res..

Edit: although seeing how much noise is generated in higher frequencies by equipments, this would translate into music when dividing and dithering i guess (unless 24 Khz is taken as broad reference for filtering)

 

[Asinus]

i find it amazing how everything above 24 Khz is left into the signal in the first place and that bigger bandwith for high-res is not used for listenable music, let's say 24 Khz.

The way i would have done it is: multiply the space for every Khz, so 24 Khz x 2 gets the space of 48 Khz, 96 Khz get's 24 Khz x 4 and so on...

so in this theory, every hertz would take double the space for storing it when recorded ( 48 Khz: 1 Hz = 2 Hz : decoder just needs to know that dividing is needed)

this way it would make sense that the bandwith of music is strecthed into bigger bandwith storage, but i guess that is not true in High res..

I'm not sure I follow what you are suggesting, but it seems like you want to pitch-shift the signal for storage which would not help with bandwidth issues.

Even when the file is a bona-fide high res recording the content above 20 kHz is almost negligible, however modern lossless codecs can reduce the size basically to the source entropy so it will encode all that empty space with way less bits and the simpler the signal the smaller the file.

 

[CRKebschull]

This xiph.org page (archived) is directly related to the topic:

24/192 Music Downloads ...and why they make no sense

...a response to the Neil Young Pono silliness... describes the human physiological reasons why any audio encoded with more than 44.1/16 resolution is not audible to human beings.

"Why push back against 24/192? Because it's a solution to a
problem that doesn't exist, a business model based on
willful ignorance and scamming people. The more that
pseudoscience goes unchecked in the world at large, the
harder it is for truth to overcome truthiness... even if
this is a small and relatively insignificant example. "

 

[voodooless]

@amirm: what would be really interesting is to only listen to the HF content. In Audition you can do that by first filtering out all output below 20 kHz (using FFT EQ), and then transposing the HF down to the audible band (using "stretch and pitch"). Potentially you'll need to raise the volume a bit as well. Then you'll actually hear what's going on there and how much actual correlation there is with the music.

The big advantage is: no issues with playing copyrighted tracks on video

 

[PO3c]

"Why push back against 24/192? Because it's a solution to a
problem that doesn't exist, a business model based on
willful ignorance and scamming people. The more that
pseudoscience goes unchecked in the world at large, the
harder it is for truth to overcome truthiness... even if
this is a small and relatively insignificant example. "

Not to defend hi-res by any means but I do have a few DVD-Audio where Bob Stuart took part in this business with his MLP https://en.wikipedia.org/wiki/Meridian_Lossless_Packing

What hi-res do not give us that lossy compression has are listening fatigue. The most common feedback now coming from listener moving from Tidal or Spotify are that they love listening to music for longer periods. So its not all bad

 

[RayDunzl]

There are some researches that hint that even if you do not hear the higher frequencies, you may still perceive them through vibrations of your organs, but as far as I know the jury is still out.

What about those of us deflicted with high frequency hearing deficiency?

As one of those myself, I can confidently report I have never noticed any sensation from playing something I can't hear.

Or do those organs (whatever that refers to) also succumb to high frequency sensatorial loss?

Maybe we don't hear with our "ears" at all.

What's the cutoff frequency for those organs?

Does a live cymbal crash tickle, or what?

Just to be safe, I say we should avoid any ultrasonic scans at the Doctor office. Don't want my spleen to go deaf.

 

[Pennyless Audiophile]

What about those of us deflicted with high frequency hearing deficiency?

As one of those myself, I can confidently report I have never noticed any sensation from playing something I can't hear.

Or do those organs (whatever that refers to) also succumb to high frequency sensatorial loss?

Maybe we don't hear with our "ears" at all.

What's the cutoff frequency for those organs?

Does a live cymbal crash tickle, or what?

Just to be safe, I say we should avoid any ultrasonic scans at the Doctor office. Don't want my spleen to go deaf.

https://www.researchgate.net/public...f_High_Resolution_Audio_Perceptual_Evaluation

 

[Maxicut]

This xiph.org page (archived) is directly related to the topic:

24/192 Music Downloads ...and why they make no sense

...a response to the Neil Young Pono silliness... describes the human physiological reasons why any audio encoded with more than 44.1/16 resolution is not audible to human beings.

"Why push back against 24/192? Because it's a solution to a
problem that doesn't exist, a business model based on
willful ignorance and scamming people. The more that
pseudoscience goes unchecked in the world at large, the
harder it is for truth to overcome truthiness... even if
this is a small and relatively insignificant example. "

It's actually a solution to a problem that used to exist

 

[martijn86]

That's the common misconception. The only thing gained by using a higher sampling rate is a higher frequency being captured, not a more accurate reconstruction.

44.1kHz samples rate gives you up to 22.05 kHz frequency response. 96 gives you up to 48, 192 gives you up to 96, etc. That's all you get. That's baked into sampling theory.

At 44.1 kHz, you are able to get everything audible, so there's no need for more. I'd like to see 48 kHz be more standard to give a little more room for proper filters, but it isn't at all necessary.

I get where he's comming from though. In practice, we all know that the audible range is covered quite nicely in 48khz. I just think that the maximum frequency is only half the sampling rate in theory and in a real world recording it wouldn't be as reliable. It is possible to generate a 24kHz sine wave on 48000 sampling points per second but in a recording you'll capture two samples in the sinewave and they aren't likely to be the exact peaks. Of course the entire discussion can be taken with a huge grain of salt because even when you have perfect hearing, there is very little musical content between 18 and 24kHz. And at the time that you can afford serious HiFi gear, your hearing will fail you after 14kHz. Don't worry too much about that. The only thing you're missing is the sound of a musquito repeller.

 

[voodooless]

It is possible to generate a 24kHz sine wave on 48000 sampling points per second but in a recording you'll capture two samples in the sinewave and they aren't likely to be the exact peaks.

Well, 24 kHz is pushing it, since you can only encode frequencies of less than half the sampling rate. In your examples, it's exactly half the sampling rate, so falls outside of the sampling theorem! But let's take 22 kHz. Your example should still hold. Looking at the samples points, you would never guess that they comprise a 22 kHz sine tone, yet after the oversampling and reconstruction filters, out comes a perfect 22 kHz sine wave. One of the many reasons to never look at staircases . If you see a picture with staircases and somebody talks about digital audio, 99% chance it's bullsh*t.

Shannon was a very smart guy! He proved that you can reconstruct a bandlimited signal by sampling it with at least double the sampling rate. More will not yield any performance benefit, other than some headroom needed for the reconstruction filter. You will also not gain any time-domain resolution. For all practical purposes, you can consider it infinite at any sampling rate for the given band limited content.

People here asserting otherwise should have a crack at disproving what Shannon proved. Good luck!

 

[sam_adams]

By the way, could you please provide some hints on how to avoid inter-sample overs? Does it require a special DAC or other hardware / software..? Or do I stray too much from the main topic?

You might take a look at this article from Benchmark on the whys and wherefores of intersample overs. I found this discussion while rummaging around in @amirm's basement. I has some other links stuck in it and I'll let you pull them out.

 

[CRKebschull]

You might take a look at this article from Benchmark on the whys and wherefores of intersample overs. I found this discussion while rummaging around in @amirm's basement. I has some other links stuck in it and I'll let you pull them out.

Oops... I think there's a link to Windows Cleartype Settings in there.

 

[sam_adams]

Oops... I think there's a link to Windows Cleartype Settings in there.

MIL! Here's the correct link:

https://benchmarkmedia.com/blogs/application_notes/intersample-overs-in-cd-recordings

 

[CRKebschull]

MIL! Here's the correct link:

https://benchmarkmedia.com/blogs/application_notes/intersample-overs-in-cd-recordings

So, on that Benchmark page, they state that intersample-overs on redbook CDs happen even though "(i)t is possible to build interpolators that will not clip or overload, but this is not being done by the D/A and SRC chip manufacturers. "

So 44.1/16 audio can deliver perfect audio which no human can hear more than it can provide. The format is not flawed for the intende use case of digital audio reproduction for human listening.

But SRC and ASRC implementations have (according to the page) technical limitations that the manufacturers just don't want to fix.

And that is why "Hi-Res Audio" is a cool thing; because it makes it easier for some manufacturers to get their D/A - A/D chain working properly? And for that you're supposed to pay extra when buying the music? But as we've seen in this thread, some Hi-Res audio files are indeed not even Hi-Res but just non-musical content in frequencies above human hearing which might infact cause audible distortions and problems for your speakers and amps?

How about a properly implemented 44.1/16 A/D - D/A chain and everyone goes home happy, like it was meant to be all along?

 

[sam_adams]

And that is why "Hi-Res Audio" is a cool thing; because it makes it easier for some manufacturers to get their D/A - A/D chain working properly? And for that you're supposed to pay extra when buying the music? But as we've seen in this thread, some Hi-Res audio files are indeed not even Hi-Res but just non-musical content in frequencies above human hearing which might infact cause audible distortions and problems for your speakers and amps?

How about a properly implemented 44.1/16 A/D - D/A chain and everyone goes home happy, like it was meant to be all along?

"Hi-Res" audio tracks can be a cool thing if you are getting actual studio masters. If not, buy a used CD and rip them yourself. You don't have to put up with "Mastered for iTunes" manipulation or any marketing jargon that is used to bilk the unknowing or unwary and you get the tracks basically as the artists/producers/engineers intended. As far as a properly implemented A/D - D/A chain, that's why there are separate DACs and why we depend on people like @amirm to bring us the tech details so we can make informed purchasing decisions.

 

[BranoO]

"Hi-Res" audio tracks can be a cool thing if you are getting actual studio masters. If not, buy a used CD and rip them yourself. You don't have to put up with "Mastered for iTunes" manipulation or any marketing jargon that is used to bilk the unknowing or unwary and you get the tracks basically as the artists/producers/engineers intended. As far as a properly implemented A/D - D/A chain, that's why there are separate DACs and why we depend on people like @amirm to bring us the tech details so we can make informed purchasing decisions.

In my opinion, the fundamental question is: Are humans able to distinguish 12kHz sine wave from 12kHz sawtooth (or any other) signal with the same level (...of what? RMS? Amplitude? 1st harmonic amplitude...)? The second harmonic is ultrasonic, so the answer is probably "no". Therefore I doubt the usefulness of hi-res audio. But it's just my opinion. I'm not trying to convince anyone.

 

[Francis Vaughan]

..of what? RMS? Amplitude? 1st harmonic amplitude...)?

Exactly the correct question. There have been many bogus proofs that there is a difference because the test did not correctly answer this question. The amplitude of the harmonic must match. The peak of a saw tooth is higher than its first harmonic, and if you don’t correctly match the levels there is an easily heard difference between the saw tooth and the sine.