Question re Hi-Res Audio

[Vladimir]

I don't care if we can't hear above 20khz, CD wasn't a perfect medium, I mean at the 11th hour Herbert Von Karajan had to convince them to extend playback time to 74 minutes. They probably should have made it 24/96 from the outset too.

It wasn't perfect, but it was damn good. I think if the engineers could do it all again with modern technology, they would have had a higher sampling frequency, just because it made some of the downstream analogue filter design easier. I dont think that they would have increased the bit depth. Absent classical music which does make use of a CD's capability, modern compressed music could probably get away with 8 or 10 bits and still have headroom to spare.

Yes. Anyone can take this test and hear for themselves how inaudible the differences between 8bits and 16bits can be.

 

[andyjm]

While you may not hear the higher frequencies, higher bit rates should yield greater detail retrieval from recordings.

+1

The answer is in the name. It's high resolution not higher frequencies.

A bit analogous with a tv picture. Think normal versus 4k, the more pixels the better the picture.

The reason 4K appears better is that current technology hasn't yet reached the full resolution of your vision. Improving the resolution of the picture above 1080p is noticable.

Redbook CD standard (16/44.1) was deliberately chosen to be better than human hearing. Higher dynamic range, and higher frequency than humans can perceive. Improving it so that it is further above your ability to perceive it - is just wasted bits.

Hi,

I would disagree. As I have posted previously, dither was added since the LSB created audible artefacts if the LSB energy was not spread within the audio band. As an analogy, the Musical Fidelity mono block has an extended frequency to 100kHz, and they state that the extended frequency has an effect which is audible. So we could either limit all amplifiers to 22kHz and there would be no impact as we cannot hear it, or try and exceed our current capabilities and see what happens. Some good may come of it. As for digital, we should always stretch the boundaries, since in the limit, a digital recording with a high sample rate and high resolution will tend towards the original analogue recording, which is what we are all pursuing.

Regards,

Shadders.

The challenge when designing equipment is to know when to stop. At the limit, everything makes a difference. You can buy analogue to digital converter ICs that work at a 10MHz sample rate for £15. So should we sample music at 10MHz? It would be nonsense to do so - you have to draw the line somewhere.

It is also wrong to consider an analogue signal as having infinite resolution, it does not. Analogue signals have a 'signal to noise ratio' in just the same way a digital signal does. The minimum signal an analogue system can resolve is its noise floor. Signals below this can not be resolved.

 

[shadders]

While you may not hear the higher frequencies, higher bit rates should yield greater detail retrieval from recordings.

+1

The answer is in the name. It's high resolution not higher frequencies.

A bit analogous with a tv picture. Think normal versus 4k, the more pixels the better the picture.

The reason 4K appears better is that current technology hasn't yet reached the full resolution of your vision. Improving the resolution of the picture above 1080p is noticable.

Redbook CD standard (16/44.1) was deliberately chosen to be better than human hearing. Higher dynamic range, and higher frequency than humans can perceive. Improving it so that it is further above your ability to perceive it - is just wasted bits.

Hi,

I would disagree. As I have posted previously, dither was added since the LSB created audible artefacts if the LSB energy was not spread within the audio band. As an analogy, the Musical Fidelity mono block has an extended frequency to 100kHz, and they state that the extended frequency has an effect which is audible. So we could either limit all amplifiers to 22kHz and there would be no impact as we cannot hear it, or try and exceed our current capabilities and see what happens. Some good may come of it. As for digital, we should always stretch the boundaries, since in the limit, a digital recording with a high sample rate and high resolution will tend towards the original analogue recording, which is what we are all pursuing.

Regards,

Shadders.

The challenge when designing equipment is to know when to stop. At the limit, everything makes a difference. You can buy analogue to digital converter ICs that work at a 10MHz sample rate for £15. So should we sample music at 10MHz? It would be nonsense to do so - you have to draw the line somewhere.

It is also wrong to consider an analogue signal as having infinite resolution, it does not. Analogue signals have a 'signal to noise ratio' in just the same way a digital signal does. The minimum signal an analogue system can resolve is its noise floor. Signals below this can not be resolved.

Hi,

I agree perhaps that for quantisation, that there will be a limit, else we are encoding the noise too. Hmmmm, maybe there might be something there. For sample rate, I think we should continue as far as possible. I seem to recall that oversampled signals at 384kHz do not sound that good compared to lower oversample rates. Why is this? Only when we push the boundaries do we discover new areas and develop new techniques. Perhaps >24bit quantisation may hold some new areas of development or research. We don't know until we try.

Regards,

Shadders.

 

[steve_1979]

We don't know until we try.

We (collectively) already have tried. Many many times.

Digital audio has been very well developed and thoroughly tested for decades. 16/44 is all you will ever need. Any analogue input can be converted to 16/44 and it will sound identical to the orginal signal. This has been tried and tested to death.

There is no need for anything higher than 16/44 unless you want to record noises that are impossible for humans to hear. I'm sorry that some people seem to be unable to understand the technical stuff which explains why this is the case. But it certainly is the case and has nothing to do with different peoples opinions. It is a simple 100% scientifically proven fact.

 

[shadders]

We don't know until we try.

We (collectively) already have tried. Many many times.

Digital audio has been very well developed and thoroughly tested for decades. 16/44 is all you will ever need. Any analogue input can be converted to 16/44 and it will sound identical to the orginal signal. This has been tried and tested to death.

There is no need for anything higher than 16/44 unless you want to record noises that are impossible for humans to hear. I'm sorry that some people seem to be unable to understand the technical stuff which explains why this is the case. But it certainly is the case and has nothing to do with different peoples opinions. It is a simple 100% scientifically proven fact.

Hi,

I would disagree. If 44.1kHz/16bit was all we needed, then why do (nearly) all commercial DAC's oversample. Why do all chipsets run at >20MHz clock frequency if all we need is a very small integer multiple of 44.1kHz?. Which proven fact states we only need 44.1kHz/16bit?.

If you referring to Nyquist, then this is based on perfect quantisation and perfect filtering of the analogue waveform. Are you therefore stating that the industry has perfect pre-filtering of the analogue waveform?

Thanks and regards,

Shadders.

 

[keeper of the quays]

I don't care if we can't hear above 20khz, CD wasn't a perfect medium, I mean at the 11th hour Herbert Von Karajan had to convince them to extend playback time to 74 minutes. They probably should have made it 24/96 from the outset too.

It wasn't perfect, but it was damn good.  I think if the engineers could do it all again with modern technology, they would have had a higher sampling frequency, just because it made some of the downstream analogue filter design easier. I dont think that they would have increased the bit depth.  Absent classical music which does make use of a CD's capability, modern compressed music could probably get away with 8 or 10 bits and still have headroom to spare. 

Yes. Anyone can take this test and hear for themselves how inaudible the differences between 8bits and 16bits can be.

 

yes i took this test..first thing i noticed that it was hardly even handed? They said before one started that there wasnt much chance of succeeding..the test is run by people who already are convinced that their right..and heres a test to prove it to the twits who disbelieve?

 

[hybridauth_Google_112658120703314018098]

Sampling at 192 or 384 khz allows for recording frequencies that are deep in the ultrasound range (and way, way out of anyone's hearing range). So yes, I don't get why that is something anyone would need or want. Would anyone be interested in a TV that output pictures not just in the visible light range but in ultra violet? It's effectively the same thing.

 

[steve_1979]

Which proven fact states we only need 44.1kHz/16bit?.

44.1kHz because nobody can hear above 22kHz

16 bit because that allows enough dynamic range to go from the queitest noise that anyone can hear right upto being so loud that it will deafen you within a few seconds all whilest also providing a noise floor (SNR) that is too low to be audible at this volume level.

This is also backed up by decades worth of listening tests too.

 

[shadders]

Which proven fact states we only need 44.1kHz/16bit?.

44.1kHz because nobody can hear above 22kHz

16 bit because that allows enough dynamic range to go from the queitest noise that anyone can hear right upto being so loud that it will deafen you within a few seconds all whilest also providing a noise floor (SNR) that is too low to be audible at this volume level.

This is also backed up by decades worth of listening tests too.

Hi,

Again I disagree. If 16bits were sufficient, then why add dither to the LSB of the recorded samples?

I believe people that state that they can hear differences, can actually hear those differences . These are usually professional people working in the industry. The fact that I cannot hear it (not sure if I can or cannot), is irrelavent, since it has been proven that some people can determine the difference. I trust those professional people that state that they can hear differences.

Regards,

Shadders.

 

[steve_1979]

... since it has been proven that some people can determine the difference. I trust those professional people that state that they can hear differences.

Who? Where?

Could you please site your sources.

 

[ID.]

This site goes on quite a bit about being able to hear differences. Also hears differences in various cables, including Ethernet. Loves him some pricey Ethernet cable.

http://www.audiostream.com/content/abx-tests-prove-hi-res-audio-legit#D4cbMuzzBdCkRwEQ.97

 

[shadders]

... since it has been proven that some people can determine the difference. I trust those professional people that state that they can hear differences.

Who? Where?

Could you please site your sources.

Hi,

Martin Colloms of Hifi Critic, and multiple people who are involved professionally in audio and contribute to the magazine.

Can you state why dither is added to the LSB of 16bit recordings if as per your statements, it cannot be heard.

Thanks and regards,

Shadders.

 

[steve_1979]

Martin Colloms of Hifi Critic, and multiple people who are involved professionally in audio and contribute to the magazine.

I'm afraid that hifi magazine reviewers don't have much credibility (apart from with a few audiophiles who don't understand how the technology works). Even if the test was performed by respected members of the scientific community the results wouldn't have much credibility unless they have been peer reviewed and the methodology of the test has been confirmed not to be flawed.

Could you site any other sources apart from a hifi magazine?

Preferably something with some credibility such as a peer reviewed scientific journal, an excerpt from a PhD thesis or research by a large reputable company such as Yamaha.

Can you state why dither is added to the LSB of 16bit recordings if as per your statements, it cannot be heard.

Dithering is used to lower the audiable noise floor even further than it otherwise would be. There's plenty of information on the net explaining how it works.

With dithering used, 16 bit audio provides enough dynamic range to reproduce the full range of volume levels which are audible to humans while simultaneously keeping the noise floor below what is audible. Even without dithering used the noise floor would still be well below what's audible in all but the most extreme of circumstances.

 

[keeper of the quays]

I used to procrastinate but now i just dither! Lol..

 

[shadders]

Martin Colloms of Hifi Critic, and multiple people who are involved professionally in audio and contribute to the magazine.

I'm afraid that hifi magazine reviewers don't have much credibility (apart from with a few audiophiles who don't understand how the technology works). Even with respected members of the scientific community the results wouldn't have much credibility unless they have been peer reviewed and the methodology of the test has been confirmed not to be flawed.

Could you site any other sources apart from a hifi magazine?

Preferably something with some credibility such as a peer reviewed scientific journal, an experpt from a PhD thesis or research by a large reputable company such as Yamaha.

Can you state why dither is added to the LSB of 16bit recordings if as per your statements, it cannot be heard.

Dithering is used to lower the audiable noise floor even further than it otherwise would be. There's plenty of information on the net explaining how it works.

With dithering used, 16 bit audio provides enough dynamic range to reproduce the full range of volume levels which are audible to humans while simultaneously keeping the noise floor below what is audible. Even without dithering used the noise floor would still be well below what's audible in all but the most extreme of circumstances.

Hi,

your comments on credibility of the people involved in Hifi Critic are not valid. Martin Colloms is a Chartered Engineer, has written a book on high performance loudspeakers, and is a well known and established loudspeaker design engineer with extensive experience. This should be sufficiently acceptable with regards to peoples ability to determine differences in bit depth of recordings.

You state that dither reduces the noise floor, do you understand the mathematics behind this, or are you repeating snippets?

You have stated that the 96dB dynamic range which 16bits resolution provides exceeds human hearing, therefore logic would indicate that dithering is not needed.

Why implement dithering if humans cannot hear it?

Unless dithering is applied, humans do register the effect of the quantisation error based on the LSB alone. Again, the entire audio industry would not use it if it was not an issue.

Regards,

Shadders.

 

[tonky]

I used to procrastinate but now i just dither! Lol..

Now I've got the jitters!

tonky

 

[steve_1979]

your comments on credibility of the people involved in Hifi Critic are not valid. Martin Colloms is a Chartered Engineer, has written a book on high performance loudspeakers, and is a well known and established loudspeaker design engineer with extensive experience. This should be sufficiently acceptable with regards to peoples ability to determine differences in bit depth of recordings.

Could you provide a link to the tests where he says that he can hear better than what 16/44.1 can provide?

You state that dither reduces the noise floor, do you understand the mathematics behind this, or are you repeating snippets?

I understand how dithering works at a theoretical level. It's quite a simple concept. At an mathematical level no I couldn't tell you exactly what the software is doing but that would require the level of knowledge and understanding of an expert and is well above what you would expect to find on a publc hifi forum.

You have stated that the 96dB dynamic range which 16bits resolution provides exceeds human hearing...

Yes when dithering is used 96dB is enough dynamic range to exceed what humans can hear.

...therefore logic would indicate that dithering is not needed.

Where did I say that dithering is not needed?

Why implement dithering if humans cannot hear it?

Unless dithering is applied, humans do register the effect of the quantisation error based on the LSB alone. Again, the entire audio industry would not use it if it was not an issue.

Exactly! I have never said that dithering doesn't help considerably. Of course it does. That's why it's used.

The fact still remains: With dithering applied 16 bit audio is enough to cover the full dynamic range that humans can hear with a low enough noise floor that it can't be heard.

 

[shadders]

Hi,

quick reply, no link stating he can hear better than 44.1kHz/16bit, but magazine reviews and forum discussions have referenced this. I accept he is being honest.

For the mathematical aspect, quoting papers out of context is not acceptable. I would expect someone that references requests for PhD papers etc., to have the capability to understand the detail and be fully conversant with the mathematics.

With regards to dithering - I am requesting that you state why it is used, and not just a statement it is used because it helps. That is, what effect is it negating for it to be of a benefit?

Regards,

Shadders.

 

[Vladimir]

I would expect someone that references requests for PhD papers etc., to have the capability to understand the detail and be fully conversant with the mathematics.

 

[Vladimir]

I use dither professionally as a graphic designer constantly, but I don't know one bit of the mathematics behind it.

I don't recall where I picked it up, but i know for ripping/converting/processing 16/44.1 I should always tick the dithered box. Is the audible benefit disputed?

 

[steve_1979]

quick reply, no link stating he can hear better than 44.1kHz/16bit, but magazine reviews and forum discussions have referenced this. I accept he is being honest.

So apart from a hifi magazine article and forum discussions, neither of which you can provide a link for, do you have any sources that show tests where humans can hear better than what 16/44 provides?

For the mathematical aspect, quoting papers out of context is not acceptable. I would expect someone that references requests for PhD papers etc., to have the capability to understand the detail and be fully conversant with the mathematics.

You don't need to understand the exact details of all the mathermatics involved. We can safely assume with a reasonably high level or confidence that academic papers such as the types I mentioned can be trusted to be correct provided that they've been peer reviewed.

You don't need to read every page and understand every bit of the mathematics involved. You only need to read the 'abstract' which will give you a brief summary of the methodology, results and conclusions. That is all the the layperson like yourself or myself needs (unless you intend to go to university and get a high enough level of education in this field to get a job in the digital audio research/design industry).

With regards to dithering - I am requesting that you state why it is used, and not just a statement it is used because it helps. That is, what effect is it negating for it to be of a benefit?

I would only be repeating what I've read on the internet. Rather than wasting time rewording what I've read would you like me to copy and paste you a description explaining how dithering works?

It's probably easier for anyone who's interested to just Google dithering and read up on it themselves like I did.

 

[Jota180]

Which proven fact states we only need 44.1kHz/16bit?.

44.1kHz because nobody can hear above 22kHz

16 bit because that allows enough dynamic range to go from the queitest noise that anyone can hear right upto being so loud that it will deafen you within a few seconds all whilest also providing a noise floor (SNR) that is too low to be audible at this volume level.

This is also backed up by decades worth of listening tests too.

Hi,

Again I disagree. If 16bits were sufficient, then why add dither to the LSB of the recorded samples?

I believe people that state that they can hear differences, can actually hear those differences . These are usually professional people working in the industry. The fact that I cannot hear it (not sure if I can or cannot), is irrelavent, since it has been proven that some people can determine the difference. I trust those professional people that state that they can hear differences.

Regards,

Shadders.

If they can hear these differences from their loudspeaker/headphones then they should also be able to use some equipment to produce a graph. They should be able to isolate this difference and play it on it's own for us all to hear.

 

[shadders]

quick reply, no link stating he can hear better than 44.1kHz/16bit, but magazine reviews and forum discussions have referenced this. I accept he is being honest.

So apart from a hifi magazine article and forum discussions, neither of which you can provide a link for, do you have any sources that show tests where humans can hear better than what 16/44 provides?

For the mathematical aspect, quoting papers out of context is not acceptable. I would expect someone that references requests for PhD papers etc., to have the capability to understand the detail and be fully conversant with the mathematics.

You don't need to understand the exact details of all the mathermatics involved. We can safely assume with a reasonably high level or confidence that academic papers such as the types I mentioned can be trusted to be correct provided that they've been peer reviewed.

You don't need to read every page and understand every bit of the mathematics involved. You only need to read the 'abstract' which will give you a brief summary of the methodology, results and conclusions. That is all the the layperson like yourself or myself needs (unless you intend to go to university and get a high enough level of education in this field to get a job in the digital audio research/design industry).

With regards to dithering - I am requesting that you state why it is used, and not just a statement it is used because it helps. That is, what effect is it negating for it to be of a benefit?

I would only be repeating what I've read on the internet. Rather than wasting time rewording what I've read would you like me to copy and paste you a description explaining how dithering works?

It's probably easier for anyone who's interested to just Google dithering and read up on it themselves like I did.

Hi,

Agreed, I do not have the links or proof that professional people in the industry can hear the difference.

With regards to understanding the mathematics, reading the abstract allows one to determine the papers details from a top level approach, but unless you understand the detail, the true meaning will not be understood. Your quotation of abstracts and admission that you do not understand the mathematics indicates that you cannot understand the impact of what you have read in the abstract.

With regards to dithering, again, you have not answered the question.

I am asking you specifically to state why dithering is used within the audio industry.

Again, if -96dB is below human hearing, and dithering affects only the LSB, why does the industry use it?

What effect is being negated by dithering?

All you are quoting is that human hearing cannot hear signal levels at -96dB based on abstracts of papers.

Please answer the question - WHY is dithering added to audio, not what the effect is of applying dithering.

Thanks and regards,

Shadders.

 

[Jota180]

Martin Colloms of Hifi Critic, and multiple people who are involved professionally in audio and contribute to the magazine.

I'm afraid that hifi magazine reviewers don't have much credibility (apart from with a few audiophiles who don't understand how the technology works). Even with respected members of the scientific community the results wouldn't have much credibility unless they have been peer reviewed and the methodology of the test has been confirmed not to be flawed.

Could you site any other sources apart from a hifi magazine?

Preferably something with some credibility such as a peer reviewed scientific journal, an experpt from a PhD thesis or research by a large reputable company such as Yamaha.

Can you state why dither is added to the LSB of 16bit recordings if as per your statements, it cannot be heard.

Dithering is used to lower the audiable noise floor even further than it otherwise would be. There's plenty of information on the net explaining how it works.

With dithering used, 16 bit audio provides enough dynamic range to reproduce the full range of volume levels which are audible to humans while simultaneously keeping the noise floor below what is audible. Even without dithering used the noise floor would still be well below what's audible in all but the most extreme of circumstances.

Hi,

your comments on credibility of the people involved in Hifi Critic are not valid. Martin Colloms is a Chartered Engineer, has written a book on high performance loudspeakers, and is a well known and established loudspeaker design engineer with extensive experience. This should be sufficiently acceptable with regards to peoples ability to determine differences in bit depth of recordings.

You state that dither reduces the noise floor, do you understand the mathematics behind this, or are you repeating snippets?

You have stated that the 96dB dynamic range which 16bits resolution provides exceeds human hearing, therefore logic would indicate that dithering is not needed.

Why implement dithering if humans cannot hear it?

Unless dithering is applied, humans do register the effect of the quantisation error based on the LSB alone. Again, the entire audio industry would not use it if it was not an issue.

Regards,

Shadders.

I don't think he's questioning the man's credibility. But he couldn't seriously hand in a HiFi article he'd written as a scientific paper. It would need to be properly set up and peer reviewed and if he's done this then I think a few would be interested to read it. If he hasn't I think people would ask why not.

I mean, Professor Prosper-Rene Blondlot was a respected professor of physics who taught at the university of Nancy and was one of the eight members of the French Academy of Sciences. He thought he discovered a new form of radiation which he called N-Rays. Blondlot, Augustin Charpentier, Arsène d'Arsonval and approximately 120 other scientists in 300 published articles claimed to be able to detect N rays emanating from most substances, including the human body with the peculiar exceptions that they were not emitted by green wood and by some treated metals.

Needless to say all these scientists were found to be seeing what they wanted to see once N-Rays were debunked by, amongst others, William Thomson, 1st Baron Kelvin.

Even though someone has a title of engineer, scientist, whatever, they need to provide peer reviewed material rather than opinion in a magazine.

 

[Jota180]

quick reply, no link stating he can hear better than 44.1kHz/16bit, but magazine reviews and forum discussions have referenced this. I accept he is being honest.

So apart from a hifi magazine article and forum discussions, neither of which you can provide a link for, do you have any sources that show tests where humans can hear better than what 16/44 provides?

For the mathematical aspect, quoting papers out of context is not acceptable. I would expect someone that references requests for PhD papers etc., to have the capability to understand the detail and be fully conversant with the mathematics.

You don't need to understand the exact details of all the mathermatics involved. We can safely assume with a reasonably high level or confidence that academic papers such as the types I mentioned can be trusted to be correct provided that they've been peer reviewed.

You don't need to read every page and understand every bit of the mathematics involved. You only need to read the 'abstract' which will give you a brief summary of the methodology, results and conclusions. That is all the the layperson like yourself or myself needs (unless you intend to go to university and get a high enough level of education in this field to get a job in the digital audio research/design industry).

With regards to dithering - I am requesting that you state why it is used, and not just a statement it is used because it helps. That is, what effect is it negating for it to be of a benefit?

I would only be repeating what I've read on the internet. Rather than wasting time rewording what I've read would you like me to copy and paste you a description explaining how dithering works?

It's probably easier for anyone who's interested to just Google dithering and read up on it themselves like I did.

Hi,

Agreed, I do not have the links or proof that professional people in the industry can hear the difference.

With regards to understanding the mathematics, reading the abstract allows one to determine the papers details from a top level approach, but unless you understand the detail, the true meaning will not be understood. Your quotation of abstracts and admission that you do not understand the mathematics indicates that you cannot understand the impact of what you have read in the abstract.

With regards to dithering, again, you have not answered the question.

I am asking you specifically to state why dithering is used within the audio industry.

Again, if -96dB is below human hearing, and dithering affects only the LSB, why does the industry use it?

What effect is being negated by dithering?

All you are quoting is that human hearing cannot hear signal levels at -96dB based on abstracts of papers.

Please answer the question - WHY is dithering added to audio, not what the effect is of applying dithering.

Thanks and regards,

Shadders.

"Whether you can hear it is another question." That, really, is the only question. Everything else is academic.