High resolution audio. The science, or lack of...?

[oldric_naubhoff]

Could I respectfully request that anyone reading/commenting about 16/44.1 CD quality audio reads about and understands Nyquist-Shannon theory first?

Nyquist-Shannon theory is used to to perfectly reproduce any analogue wave within the limits of human hearing from 16/44.1 digital information.

in general you've got two processes occuring simultaneously when you digitize an analog signal; sampling and quantizing. sampling is based on N-S theorem and it is true that sampling is error-less up to its limit, ie. sampling half band frequency. sampling in analog to digital conversion is responsible for digitazing the frequency spectrum of the signal. however, quantizing is used to capture level of the signal being digitized and quantizing, by definition, will never be error-less, simply because you've only got a limited amount of volume levels which you can apply to analog signal during quantization. as Lindsayt rightly points out; 16 bit resolution gives you 65 536 (or 2^16) discrete levels, 24 bit resolution givvevs you 16 777 216 discrete volume levels, 32 bit resolution nearly 4.3 bilion discrete volume levels. aliasing is a by-product of quantizing process and the reason why it occurs is exactly the fact that you can't fully quantize analog signal.

but on the other hand one might argue that even for 16 bit resolution the number of discrete volume levels is high enough, provided you only use a limited DR within 16bit full DR range, that you could capture the analog signal very faithfully - in effect indistinguishably from the analog input signal. still, the reality is that more bits of resolution are better for the purpose of faithfully capturing of the analog signal.

 

[steve_1979]

Read the website yet? Explains why your previous answer was way off.

Why don't the AVI pack stick to their own forum? They can't handle people even thinking questioning their wisdom, which makes for pretty stupid threads.

What really p*sses me off is the website I posted would probably backup their beliefs, not destroy them, yet they see it as a threat.

I think we have a case of crossed posts here. Read my post at the top of this page. I've seen the link that you posted but I'm not sure what your point is.

p.s. I'm not paranoid and like fr0g would be interested to see any scientific evidence/proof that disproves Nyquist-Shannon theory.

p.p.s. Let's keep it friendly and polite for the sake of the forum.

 

[Anonymous]

F$kcs sake, wrecking another? :wall:

 

[steve_1979]

For f$kcs sake, they're test tones!!! Do you think the bloody mastering will be different??

Try converting the the high rez files to 16/44.1 yourself then do the blind comparisons. They sound the same.

 

[steve_1979]

in general you've got two processes occuring simultaneously when you digitize an analog signal; sampling and quantizing. sampling is based on N-S theorem and it is true that sampling is error-less up to its limit, ie. sampling half band frequency. sampling in analog to digital conversion is responsible for digitazing the frequency spectrum of the signal. however, quantizing is used to capture level of the signal being digitized and quantizing, by definition, will never be error-less, simply because you've only got a limited amount of volume levels which you can apply to analog signal during quantization. as Lindsayt rightly points out; 16 bit resolution gives you 65 536 (or 2^16) discrete levels, 24 bit resolution givvevs you 16 777 216 discrete volume levels, 32 bit resolution nearly 4.3 bilion discrete volume levels. aliasing is a by-product of quantizing process and the reason why it occurs is exactly the fact that you can't fully quantize analog signal.

but on the other hand one might argue that even for 16 bit resolution the number of discrete volume levels is high enough, provided you only use a limited DR within 16bit full DR range, that you could capture the analog signal very faithfully - in effect indistinguishably from the analog input signal. still, the reality is that more bits of resolution are better for the purpose of faithfully capturing of the analog signal.

As far as the limits of human hearing are concerned (maximum dynamic range and maximum/minimum frequencies that we can hear) 16/44.1 resolution data can be used to perfectly reproduce any analogue wave by applying Nyquist-Shannon theory.

You would only need more data than 16/44.1 can supply if our ears could hear a greater dynamic range or a greater frequency range.

 

[altruistic.lemon]

Is it paranoia? Why can't you listen to something that may prove your pet theories? It's fear that you might be wrong? That isn't science, and neither is refusal to question established beliefs.

 

[Alec]

Is it paranoia? Why can't you listen to something that may prove your pet theories? It's fear that you might be wrong? That isn't science, and neither is refusal to question established beliefs.

Still don't know what you meant. You definitely said it, i didn't imagine it, so you may aswell explain it rather than have people make assumptions. We know what assume does, don't we?

 

[matt49]

My (very imperfect) understanding is that, as far as digitizing analogue sound is concerned, it's not the case that Nyquist-Shannon on its own guarantees perfectly faithful reproduction of the original material.

This is how I understand the theory. I have a decent scientific education, but the maths behind Nyquist-Shannon is way beyond me, as I suspect it is beyond anyone on this forum (and in that sense, asking people to 'understand' Nyquist-Shannon is a tall order), though I guess Shannon’s mathematical proof of Nyquist is irrelevant here, as I’m going to assume Nyquist-Shannon is unambiguously true.

So Nyquist-Shannon says that to reproduce a given frequency you need to sample at a rate more than twice the given frequency. If we take 20KHz as the maximum audible frequency, then you’d need to sample at more than 40KHz to reproduce the full audible frequency range. So the red book standard of 44.1KHz looks like a good choice: red book CD can faithfully reproduce everything below 22.05KHz.

However, if your original signal contains material above 22.05KHz, you’re going to get aliasing (i.e. you’ll find that two different frequencies in the material are sampled identically). By an effect known as ‘foldover’, aliasing also affects frequencies below the Nyquist frequency. To get rid of the undesirable effects of aliasing, you need to apply an anti-aliasing filter before sampling.

The problem with anti-aliasing is that the filters tend not to ‘brick-wall’ just below the Nyquist frequency, which is ideally what you’d want. Instead there’ll be a band of filter roll-off below the Nyquist frequency. In other words, there’s a risk of attenuating some of the audible frequency range. Whether or not any individual can hear anything in that range is another matter altogether.

As a result, some people think that you’re better off mastering in 24/92, which allows you to set the anti-aliasing filter a long way above the audible range.

I’ll be very happy if the assembled wisdom of the forum can demolish this.

Matt

 

[Clare Newsome]

Or we could all try listening to some music; just a thought

i've bought a few high-res albums, but only where I've read the back story behind the mastering and know it's so something someone has taken love and care to produce; often someone who loves an album and delights in the master tapes. In the best cases, where often the CDs were poorly produced to begin wth, the difference is clear. In the worst cases - like the dreadful remaster of Thriller - the difference is clear for all the wrong reasons.

The market for such remasterings - mega-bands like The Beatles aside - is only ever going to be small (especially in a world where music purchasing, rather than streaming, is in rapid decline. ) Making good money for such projects therefore means charging a premium - not an issue when that target market is typically older and relatively more affluent. Add in the high-res USP and you have your commercial proposition: you're rarely if ever going to see a CD release of these remasterings: the cost of pressing and distribution just wouldn't make it viable.

High-res music is no different to any premium product - there will be examples that buyers really feel are worth the benefit (like those who swear by a type of artisanal bread or high-thread count sheets), and others that are just about having the 'best' for sake of it.

 

[steve_1979]

Is it paranoia? Why can't you listen to something that may prove your pet theories? It's fear that you might be wrong? That isn't science, and neither is refusal to question established beliefs.

What part of "I've done the tests before" don't you understand?

But like I keep saying: For it to be a fair and valid scientific test you need to convert the high rez files to 16/44.1 yourself as this is the only way to ensure that there are no other variables which could effect the sound.

(Apologies to everyone reading for keep repeating myself but AL seems to be missing the point and keeps repeatedly asking me the same question over and over again).

 

[steve_1979]

Or we could all try listening to some music; just a thought

Don't be silly, this is a hifi forum. We don't listen to music we listen to cables and bit-rates.

 

[Clare Newsome]

Or we could all try listening to some music; just a thought

Don't be silly, this is a hifi forum. We don't listen to music we listen to cables and bit-rates.

Many a true word etc

 

[steve_1979]

Or we could all try listening to some music; just a thought

Don't be silly, this is a hifi forum. We don't listen to music we listen to cables and bit-rates.

Many a true word etc

Just pulling you leg.

You made some interesting points about the economics for why well mastered/remaster high rez audio could cost more. But there's still not reason why well mastered/remastered music it can't be distributed at 16/44.1 without losing any sound quality over high rez.

 

[steve_1979]

My (very imperfect) understanding is that, as far as digitizing analogue sound is concerned, it's not the case that Nyquist-Shannon on its own guarantees perfectly faithful reproduction of the original material.

This is how I understand the theory. I have a decent scientific education, but the maths behind Nyquist-Shannon is way beyond me, as I suspect it is beyond anyone on this forum (and in that sense, asking people to 'understand' Nyquist-Shannon is a tall order), though I guess Shannon’s mathematical proof of Nyquist is irrelevant here, as I’m going to assume Nyquist-Shannon is unambiguously true.

So Nyquist-Shannon says that to reproduce a given frequency you need to sample at a rate more than twice the given frequency. If we take 20KHz as the maximum audible frequency, then you’d need to sample at more than 40KHz to reproduce the full audible frequency range. So the red book standard of 44.1KHz looks like a good choice: red book CD can faithfully reproduce everything below 22.05KHz.

However, if your original signal contains material above 22.05KHz, you’re going to get aliasing (i.e. you’ll find that two different frequencies in the material are sampled identically). By an effect known as ‘foldover’, aliasing also affects frequencies below the Nyquist frequency. To get rid of the undesirable effects of aliasing, you need to apply an anti-aliasing filter before sampling.

The problem with anti-aliasing is that the filters tend not to ‘brick-wall’ just below the Nyquist frequency, which is ideally what you’d want. Instead there’ll be a band of filter roll-off below the Nyquist frequency. In other words, there’s a risk of attenuating some of the audible frequency range. Whether or not any individual can hear anything in that range is another matter altogether.

As a result, some people think that you’re better off mastering in 24/92, which allows you to set the anti-aliasing filter a long way above the audible range.

I’ll be very happy if the assembled wisdom of the forum can demolish this.

Matt

Good post.

IME you can't hear any anti-aliasing effects from further up the frequency range. But even if this was a concern anyone converting from high rez to 16/44.1 could just remove any information from above 22.05kHz (which is inaudible anyway) before downsampling.

By removing any infomation above 22.05kHz before downsampling, 16/44.1 is still all that is needed to perfectly reproduce any analogue wave within the dynamic range and frequency limits of any humans hearing.

 

[Clare Newsome]

Because then people wouldn't pay £20 an album, or feel they were getting a truly premium product....

When even iTunes describes itself as CD quality, that term gets devalued. 'Better than CD quality' is a powerful sell

 

[Anonymous]

Don't be silly, this is a hifi forum. We don't listen to music we listen to cables and bit-rates.

Very true, test tones, white noise and when music is played - it is probably something for the quality of the recording, not the enjoyment....

 

[steve_1979]

Because then people wouldn't pay £20 an album, so feel they were getting a truly premium product....

You may have a point there. But they would only feel that they weren't getting a truly premium product because they don't understand how the techology works. If they did understand how it works they would understand they 16/44.1 is more than enough to reproduce any analogue wave within the limits of what humans could hear.

Magazines like WHF are in a unique position to educate people that 16/44.1 is all that is needed but they still prefer to say that high rez sound better without explaining that the only reasons for the improvement is due to the mastering of the audio and that it has nothing to do with using a higher bit rate.

 

[spiny norman]

the only reasons for the improvement is due to the mastering of the audio and that it has nothing to do with using a higher bit rate.

In your not so humble opinion, of course...

 

[altruistic.lemon]

Is it paranoia? Why can't you listen to something that may prove your pet theories? It's fear that you might be wrong? That isn't science, and neither is refusal to question established beliefs.

What part of "I've done the tests before" don't you understand?

But like I keep saying: For it to be a fair and valid scientific test you need to convert the high rez files to 16/44.1 yourself as this is the only way to ensure that there are no other variables which could effect the sound.

(Apologies to everyone reading for keep repeating myself but AL seems to be missing the point and keeps repeatedly asking me the same question over and over again).

Steve, do you not understand the website, nor its aim? You claim to have read it, yet none of your replies give this impression, and repeating irrelevancies isn't helping.

Download the files, have a listen and then maybe you'll understand the point.

There's another section on the site that allows for blind testing, which would seem to be exactly what those theory-believers seem to like.

Go on, give it a go, we won't tell.

 

[steve_1979]

the only reasons for the improvement is due to the mastering of the audio and that it has nothing to do with using a higher bit rate.

In your not so humble opinion, of course...

Nothing to do with opinion. It's scientific fact.

Convert the high rez audio files to 16/44.1 and they will sound identical.

 

[altruistic.lemon]

"
Scholarly inquiry regarding scientific fact
Scholars and clinical researchers in both the social and natural sciences have forwarded numerous questions and theories in clarifying the fundamental nature of scientific fact.[22]Pertinent issues raised by this inquiry include:
the process by which "established fact" becomes recognized and accepted as such;[23]whether and to what extent "fact" and "theoretic explanation" can be considered truly independent and separable from one another;[24][25]to what extent are "facts" influenced by the mere act of observation;[25] andto what extent are factual conclusions influenced by history and consensus, rather than a strictly systematic methodology.[26][/list]
Consistent with the theory of confirmation holism, some scholars assert "fact" to be necessarily "theory-laden" to some degree. Thomas Kuhn points out that knowing what facts to measure, and how to measure them, requires the use of other theories. For example, the age of fossils is based on radiometric dating which is justified by reasoning that radioactive decay follows a Poisson process rather than a Bernoulli process. Similarly, Percy Williams Bridgman is credited with the methodological position known as operationalism, which asserts that all observations are not only influenced, but necessarily defined by the means and assumptions used to measure them.
Fact and the scientific method
Apart from the fundamental inquiry into the nature of scientific fact, there remain the practical and social considerations of how fact is investigated, established, and substantiated through the proper application of the scientific method.[27] Scientific facts are generally believed to be independent of the observer: no matter who performs a scientific experiment, all observers will agree on the outcome.[28] In addition to these considerations, there are the social and institutional measures, such as peer review and accreditation, that are intended to promote factual accuracy (among other interests) in scientific study.[29]"

So maybe not scientific fact, then?

 

[CJSF]

How to make hifi difficult, confusing and very, very boring? . . . read threads like this one.

I listed to some most enjoyable music last night, music I would not normally choose, but recommendation and an 'open mind' took me through . . . my pleasure had nothing to do with understanding science, bit rates and fancy theories.

The theories and scientific facts are another side to our hobby, they are for the boffins to fuss over so we can have some decent kit to listen to some quality recordings . . . such technicalities do not need to cloud the mind of Mr Average . . . I enjoy 'music', pure and simple, I keep that enjoyment within what I can hear and understand . . . yawn :shhh:

CJSF

 

[steve_1979]

Steve, do you not understand the website, nor its aim? You claim to have read it, yet none of your replies give this impression, and repeating irrelevancies isn't helping.

Download the files, have a listen and then maybe you'll understand the point.

There's another section on the site that allows for blind testing, which would seem to be exactly what those theory-believers seem to like.

Go on, give it a go, we won't tell.

OK this is the last time.

I. HAVE. DONE. THE. TESTS. BEFORE.

Read my reply to matt49's post above. Anti aliasing from further up the frequency range is not an issue because any audio above 22.05kHz can be removed before converting to 16/44.1 which will allow you to perfectly reproduce any analogue wave (without any anti aliasing distortion) within the dynamic range and frequency limits of human hearing.

 

[altruistic.lemon]

Steve, do you not understand the website, nor its aim? You claim to have read it, yet none of your replies give this impression, and repeating irrelevancies isn't helping.

Download the files, have a listen and then maybe you'll understand the point.

There's another section on the site that allows for blind testing, which would seem to be exactly what those theory-believers seem to like.

Go on, give it a go, we won't tell.

OK this is the last time.

I. HAVE. DONE. THE. TESTS. BEFORE.

Read my reply to matt49's post above. Anti aliasing from further up the frequency range is inaudible IME and if it worries you that much any audio above 22.05kHz can be removed before converting to 16/44.1 which will allow you to perfectly reproduce any analogue within the dynamic rangen and frequency limits of human hearing.

Well, you're a silly billy for not saying you did exactly those tests, then! Then what was all that business about re-mastering of test tones, sampling down and all that stuff? Are you worried someone else might TRY THE SAME TESTS and find a difference?

P.S. There's also a hearing test on that site. Might be worth trying, too.

 

[Richard Allen]

Can I ask a question please?.

As an ex-analogue studio engineer I have to ask " Does the bitrate really matter or is this just a discussion point?". Music by its very nature is an analogue process. In my day analogue all the way through. We then play it through an amplifier to send it through and let's face it, a rather 'iffy' medium called a loudspeaker to funnel it down our earholes.

Yes, I know I'm a luddite but I don't see all the fuss. Who cares what the bitrate is. As long as the end sound is what we want and it pleases us that's what it's all about isn't it or am I missing something here?.