24/192 a good thing?

[Alec]

For those who have a hardy constitution, here's the same debate on another forum: http://forums.linn.co.uk/bb/showthread.php?tid=15939

[/quote]

Precis of the second post - "I do not care for empirical evidence".

"...the bright ship Humana is well on her way/with brave determination/and no destination...".

 

[oldric_naubhoff]

For those who have a hardy constitution, here's the same debate on another forum: http://forums.linn.co.uk/bb/showthread.php?tid=15939

I actually made it through first 2 pages. don't want to read on (who knows, maybe I'm missing something that way). anyway, my conclusion is that most of the guys over there don't have basic knowledge about digital. they are even more clueless than me (is that even possible? ). to use some car analogy; they're like children in kindergarten arguing which car is better. the one which drives faster surely IS better!

the funniest thing is that someone mentioned an article from a Lavry engineer who refers in his article to Nyquist theorem (yes, the one on which the whole digital technology is based on), which clearly states that in order to properly recreate a wave you need to sample at rate twice as fast.

well, to tell the truth I've always thought that the higher sampling frequency the better digital technology mimics how original analog wave is shaped, therefore hi rez is better. obviously I was very much mistaken. but I guess it's not hard to get it wrong when you see something like, just for instance, this (see graph 8 and 9).

 

[CnoEvil]

I actually made it through first 2 pages. don't want to read on (who knows, maybe I'm missing something that way). anyway, my conclusion is that most of the guys over there don't have basic knowledge about digital. they are even more clueless than me (is that even possible? ). to use some car analogy; they're like children in kindergarten arguing which car is better. the one which drives faster surely IS better!

the funniest thing is that someone mentioned an article from a Lavry engineer who refers in his article to Nyquist theorem (yes, the one on which the whole digital technology is based on), which clearly states that in order to properly recreate a wave you need to sample at rate twice as fast.

well, to tell the truth I've always thought that the higher sampling frequency the better digital technology mimics how original analog wave is shaped, therefore hi rez is better. obviously I was very much mistaken. but I guess it's not hard to get it wrong when you see something like, just for instance, this (see graph 8 and 9).

It is actually worth wading through, as like here, there are a whole variety of views and opinons, with some being quite technical.

A Linn engineer gave these very technical links (way above my head):

http://www.physics.sc.edu/~kunchur/papers/FAQs.pdf http://www.physics.sc.edu/~kunchur/Acoustics-papers.htm

...much too heavy for a Sat morning.

 

[paradiziac]

Anyway, the reason this is being debated now is the media coverage of Apple reportedly being in talks with music labels to offer 24 bit downloads via the iTunes store. Apple must be under pressure from the likes of Spotify. For a small monthly fee, you can now download 320kbs files to your iPod (or streaming device) which are technically superior to Apple's 256kbs. Apple can't offer 16/44 lossless because it would have to be priced above their lossy offering. If they did that, people would no doubt realize that it would be cheaper to order the CD from Amazon and rip it, they could even throw it in the bin afterwards!So...along comes a 24bit, "better than CD", "HD" version on iTunes that takes up way more space (conveniently needing an iDevice upgrade due to greater storage demands), costs more and...sounds the same (though people might just "hear a difference"). Any takers?
Anyway, the reason this is being debated now is the media coverage of Apple reportedly being in talks with music labels to offer 24 bit downloads via the iTunes store.

Apple must be under pressure from the likes of Spotify. For a small monthly fee, you can now download 320kbs files to your iPod (or streaming device) which are technically superior to Apple's 256kbs. Apple can't offer 16/44 lossless because it would have to be priced above their lossy offering. If they did that, people would no doubt realize that it would be cheaper to order the CD from Amazon and rip it, they could even throw it in the bin afterwards!

So...along comes a 24bit, "better than CD", "HD" version on iTunes that takes up way more space (conveniently needing an iDevice upgrade due to greater storage demands), costs more and...sounds the same (though people might just "hear a difference").

Makes sense. For Apple, if they can pull that one off, that is.

 

[paradiziac]

Just found this:

http://www.apple.com/itunes/inside-itunes/2012/02/mastered-for-itunes-means-superior-digital-sound-quality.html

So they are persuading labels to re-master stuff which they will no doubt later release as "hi-res" 24 bit files and charge everyone to upgrade their collections.

If you want a good master, you'll have to buy it from Apple and it helps if you play it on an Apple device...

I bet they will even get away with selling lossy 24 bit files.

It will sound better, but not because it's "hi-res"...but no-one will understand or care.

Clever.

 

[CnoEvil]

Anyway, the reason this is being debated now is the media coverage of Apple reportedly being in talks with music labels to offer 24 bit downloads via the iTunes store. Apple must be under pressure from the likes of Spotify. For a small monthly fee, you can now download 320kbs files to your iPod (or streaming device) which are technically superior to Apple's 256kbs. Apple can't offer 16/44 lossless because it would have to be priced above their lossy offering. If they did that, people would no doubt realize that it would be cheaper to order the CD from Amazon and rip it, they could even throw it in the bin afterwards!So...along comes a 24bit, "better than CD", "HD" version on iTunes that takes up way more space (conveniently needing an iDevice upgrade due to greater storage demands), costs more and...sounds the same (though people might just "hear a difference"). Any takers?
Anyway, the reason this is being debated now is the media coverage of Apple reportedly being in talks with music labels to offer 24 bit downloads via the iTunes store.

Apple must be under pressure from the likes of Spotify. For a small monthly fee, you can now download 320kbs files to your iPod (or streaming device) which are technically superior to Apple's 256kbs. Apple can't offer 16/44 lossless because it would have to be priced above their lossy offering. If they did that, people would no doubt realize that it would be cheaper to order the CD from Amazon and rip it, they could even throw it in the bin afterwards!

So...along comes a 24bit, "better than CD", "HD" version on iTunes that takes up way more space (conveniently needing an iDevice upgrade due to greater storage demands), costs more and...sounds the same (though people might just "hear a difference").

Makes sense. For Apple, if they can pull that one off, that is.

You are probably right.

There is always an agenda, with two sides, both backed up by irrefutable science.

 

[John Duncan]

There is always an agenda, with two sides, both backed up by irrefutable science.

I might set up a forum with that very title. Or possibly just "Somebody is wrong on the internet".

 

[Clare Newsome]

We really should get t-shirts/mugs made up...

 

[CnoEvil]

There is always an agenda, with two sides, both backed up by irrefutable science.

I might set up a forum with that very title. Or possibly just "Somebody is wrong on the internet".

THE TRUTH IS OUT THERE!

 

[Overdose]

In my view, all downloads should be priced according to file size, with a minimum fee set for the lowest bit rates. I think that if you are charging CD money for a download it should be CD quality and have all the associated extras, such as cover art and inlay documents.

The price premium for higher bitrates should relate directly to storage costs and not be arbritrarily high, as is the case for 'high res' files. A Russian site 'Allofmp3.com' used to use this similar pricing structure, although it later became apparent that the site was illegal. It was however, very well laid out and implemented and the same sort of model would do very well if legitimised.

At any rate, if true high resolution files become ubiquitous and reasonably priced, they make the obvious choice for archiving and can be compressed later according to the type of playback device being used. Regardless of whether or not there is any audible difference between 16 and 24 bit versions of the same master, you can be sure that you will find no better quality than the 24bit version.

 

[shadders]

For those who have a hardy constitution, here's the same debate on another forum: http://forums.linn.co.uk/bb/showthread.php?tid=15939

I actually made it through first 2 pages. don't want to read on (who knows, maybe I'm missing something that way). anyway, my conclusion is that most of the guys over there don't have basic knowledge about digital. they are even more clueless than me (is that even possible? ). to use some car analogy; they're like children in kindergarten arguing which car is better. the one which drives faster surely IS better!

the funniest thing is that someone mentioned an article from a Lavry engineer who refers in his article to Nyquist theorem (yes, the one on which the whole digital technology is based on), which clearly states that in order to properly recreate a wave you need to sample at rate twice as fast.

well, to tell the truth I've always thought that the higher sampling frequency the better digital technology mimics how original analog wave is shaped, therefore hi rez is better. obviously I was very much mistaken. but I guess it's not hard to get it wrong when you see something like, just for instance, this (see graph 8 and 9).

Hi,

No you are not wrong. The sampling theorem states the sampling rate must be at least 2 x max frequency (greater than or equal to) you are trying to sample for a band limited signal. This is theory - since if you sampled a 20kHz sine wave with a 40kHz sample frquency, and those samples were at the zero crossing point, then there would be no data. If you sample at the peak of the sine wave - then ok, your reconstruction filter will be able to approximate a sine wave.

As such, the greater the sampling frequency and the more bits per sample, the less quantisation error introduced into the reconstructed signal, and hence better quality.

If you took this to the limit, 100 bits per sample (example), and sampled every nano second, you would in effect be reconstructing an analogue signal nearly perfectly.

Regards,

Shadders.

 

[CnoEvil]

Hi,

No you are not wrong. The sampling theorem states the sampling rate must be at least 2 x max frequency (greater than or equal to) you are trying to sample for a band limited signal. This is theory - since if you sampled a 20kHz sine wave with a 40kHz sample frquency, and those samples were at the zero crossing point, then there would be no data. If you sample at the peak of the sine wave - then ok, your reconstruction filter will be able to approximate a sine wave.

As such, the greater the sampling frequency and the more bits per sample, the less quantisation error introduced into the reconstructed signal, and hence better quality.

If you took this to the limit, 100 bits per sample (example), and sampled every nano second, you would in effect be reconstructing an analogue signal nearly perfectly.

Regards,

Shadders.

I tried to argue (rather simplistically) the same thing here:
http://www.whathifi.com/forum/computer-based-music/mp3-320-kbps-vs-flacwav?page=5

...but was given reasonable logic why this wasn't so; but since I can hear the difference, I think I'm right (but who knows). :?

 

[shadders]

Hi,

No you are not wrong. The sampling theorem states the sampling rate must be at least 2 x max frequency (greater than or equal to) you are trying to sample for a band limited signal. This is theory - since if you sampled a 20kHz sine wave with a 40kHz sample frquency, and those samples were at the zero crossing point, then there would be no data. If you sample at the peak of the sine wave - then ok, your reconstruction filter will be able to approximate a sine wave.

As such, the greater the sampling frequency and the more bits per sample, the less quantisation error introduced into the reconstructed signal, and hence better quality.

If you took this to the limit, 100 bits per sample (example), and sampled every nano second, you would in effect be reconstructing an analogue signal nearly perfectly.

Regards,

Shadders.

I tried to argue (rather simplistically) the same thing here: http://www.whathifi.com/forum/computer-based-music/mp3-320-kbps-vs-flacwav?page=5 ...but was given reasonable logic why this wasn't so; but since I can hear the difference, I think I'm right (but who knows). :?

Hi

The quote on the page referenced ""in theory" is very valid. But then 24/96 files do not give you any more resolution. They use the same algorithm. There is no filling in, just extra Hz above what we can hear, and extra potential dynamic range"

Is incorrect. The 24bit sample allows more granularity of the voltage signal being sampled - as such - there is less quantisation error (noise). The higher the sampling rate allows for improved accuracy when reconstructing the signal.

You are correct in your understanding, more bits and higher sampling frequency will allow for a higher quality signal. As the quote states - you will get an higher frequency bandwidth of the signal - correct, , but the other statements in the quote are wrong.

Regards,

Shadders.

 

[CnoEvil]

Hi

The quote on the page referenced ""in theory" is very valid. But then 24/96 files do not give you any more resolution. They use the same algorithm. There is no filling in, just extra Hz above what we can hear, and extra potential dynamic range"

Is incorrect. The 24bit sample allows more granularity of the voltage signal being sampled - as such - there is less quantisation error (noise). The higher the sampling rate allows for improved accuracy when reconstructing the signal.

You are correct in your understanding, more bits and higher sampling frequency will allow for a higher quality signal. As the quote states - you will get an higher frequency bandwidth of the signal - correct, , but the other statements in the quote are wrong.

Regards,

Shadders.

Thank you for the verification.

 

[Anonymous]

well, to tell the truth I've always thought that the higher sampling frequency the better digital technology mimics how original analog wave is shaped, therefore hi rez is better. obviously I was very much mistaken. but I guess it's not hard to get it wrong when you see something like, just for instance, this (see graph 8 and 9).

What you're looking at in figures 8 and 9 isn't 44/48 vs 96 kHz sampling, it's 16-bit vs 24-bit amplitude resolution, using a signal that is so quiet that it can hardly be represented in 16 bit (sound is at -90dB, whereas 16 bit has a total range of 96dB). It's no surprise then that 24 bit is far more accurate, since it can use 10 bits to represent the waveform as opposed to 2 bits.

(yes I know, I promised not to repond. Just adding some clarification)

 

[SteveR750]

and don't forget dither. Ah, only if you are recording and mixing. More bits must be helpful, if only because any digital processing could take place without messing with the SQ after conversion.

 

[steve_1979]

We really should get t-shirts/mugs made up...

+1

 

[WinterRacer]

As such, the greater the sampling frequency and the more bits per sample, the less quantisation error introduced into the reconstructed signal, and hence better quality.

Shadders, remember that dither will turn the quantisation errors into random noise well below the noise floor that you can hear. All that 24bits give you is an increased dynamic range that the rest of your equipment can't reproduce and even if it could your ears couldn't take.

A higher sampling rate will just allow reproduce sounds you can't hear, or if so loud you cold hear, they'd be painful.

High sampling rates and bit depths have a place in the recording studio, but completely pointless for the rest of us (IMHO )

 

[shadders]

As such, the greater the sampling frequency and the more bits per sample, the less quantisation error introduced into the reconstructed signal, and hence better quality.

Shadders, remember that dither will turn the quantisation errors into random noise well below the noise floor that you can hear. All that 24bits give you is an increased dynamic range that the rest of your equipment can't reproduce and even if it could your ears couldn't take.

A higher sampling rate will just allow reproduce sounds you can't hear, or if so loud you cold hear, they'd be painful.

High sampling rates and bit depths have a place in the recording studio, but completely pointless for the rest of us (IMHO )

Hi,

Dither is an effect used to compensate for the lack of bits. Article following indicates this too :

http://www.earlevel.com/main/1996/10/20/what-is-dither/

All dither does as you have said is to modulate the least significant bit to be less discernable to the ear with regards to quantisation noise, and if modulated correctly can provide the listener with the effect that they can hear lower signals than the 16bit limit.

Dither was applied since the 16bit format was recognised as not being adequate, but technology for CD was established to be 16bit.

A higher sampling rate will allow intermediate steps between the usual 44.1kHz sampling frequency to be recorded, thus providing an greater level of accuracy on the reproduced waveform for the filter to smooth out.

Dither as you have stated was to remove the effect of quantisation noise - but it is only an effect.

Higher sampling rates and greater sampling word size (bits) will provide a more accurate reconstructed signal and the use of dither will not be required.

Regards,

Shadders.