Hi resolution audio and extended treble response!

[dogmamann]

I undertand hi res audio is Not something which would have an audible difference. Let’s ignore that now, and can anyone explain why Sonys website has two diagrams contradicting to each other here?

Above shows hi resolution means you have some information added above 20khz!

Below one shows how the audible range itself is sampled 192 times, instead of 44 samples.

Out of the two what exactly is the theoretical hi res?

 

[Blumlein 88]

I undertand hi res audio is Not something which would have an audible difference. Let’s ignore that now, and can anyone explain why Sonys website has two diagrams contradicting to each other here?

Above shows hi resolution means you have some information added above 20khz!

Below one shows how the audible range itself is sampled 192 times, instead of 44 samples.

Out of the two what exactly is the theoretical hi res?

Is something missing? I only see one image so not sure what the comparison is.

In general higher sample rates allow higher frequency response. According to Shannon's work you only need 2 or more samples to completely reconstruct a sampled waveform. Constructing from 4 or 8 samples doesn't improve accuracy of the reconstruction. It would only allow it continue at a higher frequency.

 

[dogmamann]

Is something missing? I only see one image so not sure what the comparison is.

In general higher sample rates allow higher frequency response. According to Shannon's work you only need 2 or more samples to completely reconstruct a sampled waveform. Constructing from 4 or 8 samples doesn't improve accuracy of the reconstruction. It would only allow it continue at a higher frequency.

One image and the image in the video

 

[dogmamann]

We cannot sample 20 to 20khz more than 44khz?

 

[Blumlein 88]

We cannot sample 20 to 20khz more than 44khz?

Yes you can, but it contributes nothing to the accuracy of the signals below 20 khz. The higher sample rates only extend the response to higher frequencies. For example maybe you have a tuning fork at 4096 hz. It will be sampled and reconstructed accurately at 44 khz sampling rates. It will be sampled and reconstructed accurately at 192 khz sampling rates, but the latter is not more accurate than the lower rate.

 

[BDWoody]

We cannot sample 20 to 20khz more than 44khz?

It does seem like more samples would give a more accurate result, but the magic is that after you've gotten your 2 samples, more doesn't increase fidelity for that given frequency. Higher sample rates make it so you can get those two samples from higher and higher frequencies (88kHz sample rate let's you recreate frequencies up to 44 kHz, 192kHz sample rate up to 96kHz, etc) but that's really all you are going to get.

 

[dogmamann]

It does seem like more samples would give a more accurate result, but the magic is that after you've gotten your 2 samples, more doesn't increase fidelity for that given frequency. Higher sample rates make it so you can get those two samples from higher and higher frequencies (88kHz sample rate let's you recreate frequencies up to 44 kHz, 192kHz sample rate up to 96kHz, etc) but that's really all you are going to get.

Ok, so for 20hz signal 40 samples is the max sample rate for audible fidelity.? We cannot sample 20hz in more than 40 samples ? I am not taking about the audibility of it, but is it possible ?

 

[BDWoody]

Ok, so for 20hz signal 40 samples is the max sample rate for audible fidelity.? We cannot sample 20hz in more than 40 samples ? I am not taking about the audibility of it, but is it possible ?

You certainly can, but unless you ONLY want 20Hz, you will need more than a 40Hz sample rate. It will get sampled at the overall rate (say 44.1kHz) , and there will be lots of extra samples for lower frequencies, with fewer and fewer as the frequencies rise up to the Nyquist frequency, which is going to be exactly 1/2 the sample rate (because that's the highest frequency that can still get 2 samples squeezed in) of 22.05 kHz.

The extra samples don't help or hurt, any more than me taking more measurements of the diameter of a circle will help me recreate it more or less accurately.

 

[jsilvela]

You certainly can, but unless you ONLY want 20Hz, you will need more than a 40Hz sample rate. It will get sampled at the overall rate (say 44.1kHz) , and there will be lots of extra samples for lower frequencies, with fewer and fewer as the frequencies rise up to the Nyquist frequency, which is going to be exactly 1/2 the sample rate (because that's the highest frequency that can still get 2 samples squeezed in) of 22.05 kHz.

The extra samples don't help or hurt, any more than me taking more measurements of the diameter of a circle will help me recreate it more or less accurately.

I took signal processing ages ago. I remember the awe learning about the sampling theorem.
I've often wondered why the sampling rate is 44.1kHz.
I assume it's not 40kHz to give some tolerance to the low pass filters so they don't need to be a brick wall.

Is the extra "cushion" enough? Are there crappy electronics where the low pass filters don't fall fast enough, and produce aliasing?
Might a higher sampling rate have resulted in easier/cheaper filters?

 

[jsilvela]

I took signal processing ages ago. I remember the awe learning about the sampling theorem.
I've often wondered why the sampling rate is 44.1kHz.
I assume it's not 40kHz to give some tolerance to the low pass filters so they don't need to be a brick wall.

Is the extra "cushion" enough? Are there crappy electronics where the low pass filters don't fall fast enough, and produce aliasing?
Might a higher sampling rate have resulted in easier/cheaper filters?

Answering myself:

44,100 Hz - Wikipedia

en.wikipedia.org

The Nyquist–Shannon sampling theorem says the sampling frequency must be greater than twice the maximum frequency one wishes to reproduce. Since human hearing range is roughly 20 Hz to 20,000 Hz, the sampling rate had to be greater than 40 kHz.

In addition, signals must be low-pass filtered before sampling to avoid aliasing. While an ideal low-pass filter would perfectly pass frequencies below 20 kHz (without attenuating them) and perfectly cut off frequencies above 20 kHz, such an ideal filter is theoretically and practically impossible to implement as it is noncausal, so in practice a transition band is necessary, where frequencies are partly attenuated. The wider this transition band is, the easier and more economical it is to make an anti-aliasing filter. The 44.1 kHz sampling frequency allows for a 2.05 kHz transition band.

 

[antcollinet]

One image and the image in the video

Which video?

 

[dogmamann]

Which video?

 

[DVDdoug]

Above shows hi resolution means you have some information added above 20khz!

You CAN have recordings with higher frequency content. Sometimes there is some ultrasonic noise even if the program material doesn't extend beyond the audio range. Microphones have a limited range and most preamps and any other analog electronics don't extend much beyond the audio range.

We cannot sample 20 to 20khz more than 44khz?...

...Ok, so for 20hz signal 40 samples is the max sample rate for audible fidelity.? We cannot sample 20hz in more than 40 samples ?

There's no theoretical upper-limit to how fast you can sample. Digital audio files can contain zero-Hz "information" (AKA a "DC offset")

The easiest way to understand the theory is that you need at least one sample for the positive-half of the wave and one sample for the bottom half. If you sample less frequently you get aliasing ("false frequencies" below the Nyquist limit).

So that's an absolute limit.

I believe the "perfect reconstruction" part of the theory requires a constant signal of infinite duration so many cycles every-part of the wave gets sampled. And you can get some "interesting results" (modulation) if you generate a 22,049Hz signal at 44.1kHz in Audacity. You'll also get visible (and possibly audible) modulation 11,024 or 11026Hz, but not as much.

Luckily, our hearing isn't perfect either!

 

[Jimbob54]

You need to link to that video. As it stands you have provided a screenshot- we need the link please

EDIT- I dun some digging- its available in this page https://www.sony.co.uk/electronics/hi-res-audio

Basic premise- higher than 16bit allows greater DR
Higher than 44.1khz sample rate allows for higher freq than 20khz to be reproduced - "making it more lifelike"- its the last bit that I think many here would take issue with.

 

[JustJones]

I watched the video. I think what the Sony marketing guy was saying is since audio has been recorded in higher resolution for years at 24/96 then obviously we'll get a better
" emotional response " listening in higher resolution of 24/96 so buy our stuff that will play higher resolution. The 2 photos shown above are taken from different places on the same page. One is a drawing the guy does in the video the other is further down on the page, I think they're basically trying to infer the same thing.

 

[dogmamann]

So that means half of our speakers cannot theoretically produce hi res ! When I look at the response of many of them, many stops or have a steep roll off after 20khz! So those according to Sonys logic are not hi res capable !

 

[dogmamann]

I am still not sure if hi res means:

Files with frequencies above 20khz upto 192khz
(Main waves + its harmonics in high frequency region)

Or

Files with lot more samples representing the content the recording having 20 to 20khz

 

[Apesbrain]

So those according to Sonys logic are not hi res capable !

My ears are not "hi res" capable.

 

[antcollinet]

My ears are not "hi res" capable.

True.

Further, nor are anyone else's.

 

[antcollinet]

I am still not sure if hi res means:

Files with frequencies above 20khz upto 192khz
(Main waves + its harmonics in high frequency region)

Or

Files with lot more samples representing the content the recording having 20 to 20khz

It is the first one (But up to a bit below half the sample rate - so around 90Khz for the 192kHz sample rate). But - Hi Res means delivering frequencies higher than the human ear can hear. It is basically pointless.