High Resolution Audio?

[Calexico]

Indeed. That's how oversampling works. New values between the already present samples are calculated. No higher frequencies in between the samples are generated/recovered/invented.
The only benefit it can have is mentioned in my earlier post. When poorer or no filters are used the filter frequency of the DAC itself is higher and doesn't affect the audible range any more.
Also DAC's with a gentle filter that would show massive amounts of energy above 22kHz would 'shift' that garbage (as the upsampling algo usually has a sharp filter) moves all the garbage that DAC produces to above the nyquist of the upsampled frequency or in the case of DSD into the rising noise.

When we see that the filter are -10db at 22khz on the dac does it mean that there will be noise just after 22khz? ( Attenuated by only 10 db?)

 

[eliash]

When we see that the filter are -10db at 22khz on the dac does it mean that there will be noise just after 22khz? ( Attenuated by only 10 db?)

What I don´t understand about the upsampling discussion is, why to do it before, when the DAC-chip does it anyway, in its final oversampling single or multibit D/A block. The overshoot above 0dBfs of the analog signal output proves that, when applying a clipped digital input!?

 

[solderdude]

I think folks upsample because of afformentioned reasons or to bypass the internal reconstruction filter (believing it is not good enough) or they upsample because they are told it sounds better.

What garbage is present above 20kHz has been discussed at length (look for replies from mansr for instance)

 

[maty]

The world beyond 20kHz

[PDF] https://www.earthworksaudio.com/wp-content/uploads/2012/07/The-world-beyond-20kHz.pdf

[ THERE IS MUCH controversy about how we might move forward towards higher quality reproduction of sound. The compact-disc standard assumes that there is no useful information beyond 20kHz and therefore includes a brick-wall filter just above 20kHz. Many listeners hear a great difference when 20kHz band-limited audio signals are compared with wide band signals. A number of digital systems have been proposed which sample audio signals at 96kHz and above, and with up to 24 bits of quantization.

Many engineers have been trained to believe that human hearing receives no meaningful input from frequency components above 20kHz. I have read many irate letters from such engineers insisting that information above 20kHz is clearly useless, and any attempts to include such information in audio signals is deceptive, wasteful and foolish, and that any right-minded audio engineer should realize that this 20kHz limitation has been known to be an absolute limitation for many decades. Those of us who are convinced that there is critically important audio information to at least 40kHz are viewed as misguided... ]

 

[solderdude]

Maty.... What was discussed here is NOT hires but the benefits of up-sampling.
Arguably any slow filter that produces signals above Nyquist (talking 44/16 here) has absolutely NO audio information, in fact these are signals that are 'mirrored' and have no harmonic relation to sound and can potentially do more harm than not.
In THIS case there is absolutely NO benefit at all from having any signal there.

 

[edechamps]

Many listeners hear a great difference when 20kHz band-limited audio signals are compared with wide band signals

According to arguably the best evidence we have on the subject, that is simply not true. A more accurate statement would be:

Some highly trained listeners can hear a very slight difference when 20kHz band-limited audio signals are compared with wide band signals

 

[SIY]

According to arguably the best evidence we have on the subject, that is simply not true. A more accurate statement would be:

"Some highly trained listeners can hear a very slight difference when 20kHz band-limited audio signals are compared with wide band signals with certain test material."

 

[Blumlein 88]

"Some highly trained listeners can hear a very slight difference when 20kHz band-limited audio signals are compared with wide band signals with certain test material."

"Some highly trained listeners can hear a very slight difference when 20kHz band-limited audio signals are compared with wide band signals with certain test material. Under extraordinarily good listening conditions using exceptionally good systems of reproduction with extended response to 40 khz"

 

[SIY]

Well, things are narrowing down nicely.

 

[Blumlein 88]

The world beyond 20kHz

[PDF] https://www.earthworksaudio.com/wp-content/uploads/2012/07/The-world-beyond-20kHz.pdf

[ THERE IS MUCH controversy about how we might move forward towards higher quality reproduction of sound. The compact-disc standard assumes that there is no useful information beyond 20kHz and therefore includes a brick-wall filter just above 20kHz. Many listeners hear a great difference when 20kHz band-limited audio signals are compared with wide band signals. A number of digital systems have been proposed which sample audio signals at 96kHz and above, and with up to 24 bits of quantization.

Many engineers have been trained to believe that human hearing receives no meaningful input from frequency components above 20kHz. I have read many irate letters from such engineers insisting that information above 20kHz is clearly useless, and any attempts to include such information in audio signals is deceptive, wasteful and foolish, and that any right-minded audio engineer should realize that this 20kHz limitation has been known to be an absolute limitation for many decades. Those of us who are convinced that there is critically important audio information to at least 40kHz are viewed as misguided... ]

Unfortunately he is making the same old mistake. Thinking that timing accuracy is limited to sampling rate. Also thinking that timing accuracy is limited by frequency response. Both are simply wrong.

Earthworks does make excellent microphones. I've one of theirs and when I can use it I would say it is my finest sounding microphone.

 

[mansr]

Well, things are narrowing down nicely.

 

[SIY]

Is the unit of measurement the RCH?

 

[Blumlein 88]

When we see that the filter are -10db at 22khz on the dac does it mean that there will be noise just after 22khz? ( Attenuated by only 10 db?)

Here is an FFT of a DAC running at 48 khz reproducing -4 db white noise. It was recorded at 192 khz so you can see what happens above nyquist or in this case 24 khz. This lets you see the shape of the imaging filter in the DAC. This one uses a half band filter. Meaning instead of being down 96 db at 24 khz it is only down about 48 db at 24 khz. It continues to drop off being down to the basic analog noise floor by a little over 26 khz.

In your example of a DAC only down -10 db at 22 khz (presumably running at 44.1 khz) you ask what happens just after 22 khz. The answer is depends upon the filter. If it were a very slow roll off, not reducing levels to -96 db until 40 khz, then a white noise signal would bleed over into a sloped response only being filtered out around 40 khz. Meaning some decreasing amounts of noise up above 20 khz.

Here below is the same DAC fed a high level tone near 20 khz. You see the tone, you see a low level of 2nd and 3rd harmonic distortion. You see imaging artifacts at 28 and 52 khz. The others are from various other sources just ignore them. The 28 khz image is at -116 db because of the filter. And the 52 khz mirror image is -97 db.

In your example such a DAC with a slow roll off filter the 28 khz tone might be only -12 db down in level and the 52 khz might be only slightly lower than that.

NOTE:my frequencies may appear a little off from the label, but that is because this was captured during a sweep. So the tone was really more like 19,500 hz instead of 20 khz. Close enough to illustrate the point I hope.

 

[tohoho4]

"Some highly trained listeners can hear a very slight difference when 20kHz band-limited audio signals are compared with wide band signals with certain test material."

Most ordinary listeners can't hear the difference between 20kHz band-limited audio signals and wide band signals, but wide band signals produce a physiological effect on brain activity accroding to the following article.

https://www.physiology.org/doi/full...d=ori:rid:crossref.org&rfr_dat=cr_pub=pubmed&

 

[Blumlein 88]

Most ordinary listeners can't hear the difference between 20kHz band-limited audio signals and wide band signals, but wide band signals produce a physiological effect on brain activity accroding to the following article.

https://www.physiology.org/doi/full/10.1152/jn.2000.83.6.3548?url_ver=Z39.88-2003&rfr_id=ori:rid:crossref.org&rfr_dat=cr_pub=pubmed&

I believe, but may be mixing up some of Oohashi's tests, that this one was done by two other researchers and they failed to replicate these results.

 

[Ron Party]

Didn't Oohashi's test get debunked?

 

[Blumlein 88]

Here is a good summary around the matter.

http://www.hificritic.com/uploads/2...c-sc10-do_we_need_an_ultrasonic_bandwidth.pdf

 

[SIY]

Didn't Oohashi's test get debunked?

That poor old horse gets trotted out regularly despite being highly lame.

 

[Calexico]

Here is an FFT of a DAC running at 48 khz reproducing -4 db white noise. It was recorded at 192 khz so you can see what happens above nyquist or in this case 24 khz. This lets you see the shape of the imaging filter in the DAC. This one uses a half band filter. Meaning instead of being down 96 db at 24 khz it is only down about 48 db at 24 khz. It continues to drop off being down to the basic analog noise floor by a little over 26 khz.


View attachment 27656

In your example of a DAC only down -10 db at 22 khz (presumably running at 44.1 khz) you ask what happens just after 22 khz. The answer is depends upon the filter. If it were a very slow roll off, not reducing levels to -96 db until 40 khz, then a white noise signal would bleed over into a sloped response only being filtered out around 40 khz. Meaning some decreasing amounts of noise up above 20 khz.

Here below is the same DAC fed a high level tone near 20 khz. You see the tone, you see a low level of 2nd and 3rd harmonic distortion. You see imaging artifacts at 28 and 52 khz. The others are from various other sources just ignore them. The 28 khz image is at -116 db because of the filter. And the 52 khz mirror image is -97 db.
View attachment 27659

In your example such a DAC with a slow roll off filter the 28 khz tone might be only -12 db down in level and the 52 khz might be only slightly lower than that.

NOTE:my frequencies may appear a little off from the label, but that is because this was captured during a sweep. So the tone was really more like 19,500 hz instead of 20 khz. Close enough to illustrate the point I hope.

With sharp filter at -10db at 22.5khz (as we see on some tests here)
Are there frequencies which give artefacts just after nyquist and then that are lowered just by 10db?
Is it better a sharp filter that would be -96db just after nyquist? Are there dacs that do it?

 

[GrimSurfer]

That poor old horse gets trotted out regularly despite being highly lame.

If Oohashi is a true scientist, he would be just as pleased with having his theories disproven as proven.

The real take away that I draw from the paper is this: Commericial interest is both has both a limiting and enabling influence on bullshit.

Those in the early radio business would have profited from setting a low range of human hearing. Microphone, transmitter, recorder and loudspeaker costs would have been reduced if, say, the accepted range was 20 Hz to 10 kHz.

Those in the contemporary audio business can, and often do, profit by advancing the postulates that human hearing and perception continue past ~18 kHz. Doing so opens all kinds of commercial opportunities regardless of the legitimacy of the underlying science.

All of that said, it is noteworthy the BBC (a publicly funded Institution that operates a multitude of "for profit" services) seems to have arrived at the closest, earliest definition of audible sound. It would be interesting to learn more about the corporate culture and workforce in place that allowed (or enabled) this to happen, as it might reveal a model that legitimate audio companies could use.

Where there's a buck to be saved or made, people will try to shape outcomes accordingly. That piece may never change!