How can CDs be lossless?

[inbox4]

I've been reading about streaming and the different file formats recently.

I keep reading that CDs are lossless.

Could someone explain this to me please? If CDs are lossless at 16/44.1, why do higher res recordings exist?

I dont want to get into the discussion of whether you can hear the difference of the high res formats over CD. However, I would like to understand why if the 24 bit recordings contain more information, how can the CD be considered lossless if by comparison data is missing?

- I hope that makes sense.

Thanks.

 

[Paul.]

I've been reading about streaming and the different file formats recently.

I keep reading that CDs are lossless.

Could someone explain this to me please? If CDs are lossless at 16/44.1, why do higher res recordings exist?

I dont want to get into the discussion of whether you can hear the difference of the high res formats over CD. However, I would like to understand why if the 24 bit recordings contain more information, how can the CD be considered lossless if by comparison data is missing?

- I hope that makes sense.

Thanks.

CDs are not lossless. Lossless refers to a rip of a CD generally. A lossless rip of a CD means all data from the CD is either present (i.e. WAV), or present after decompression (I.e. FLAC, ALAC).

 

[spiny norman]

CDs are not lossless. Lossless refers to a rip of a CD generally. A lossless rip of a CD means all data from the CD is either present (i.e. WAV), or present after decompression (I.e. FLAC, ALAC).

Exactly: as opposed to lossy compression formats such as MP3 or AAC, which throw away some of the data irretrivably in order to reduce file sizes, lossless formats work like a .zip file, in that all the data is retained.

eg If you could use FLAC for this sentence 'Using compression to reduce the total file size, some of the less useful words used in it may perhaps be lost along the way', when uncompressed again all the words would still be there.

With MP3 you might end up with something like 'Using compression to reduce file size, some words may be lost.'

The sentence still makes sense and carries the meaning, but some detail is missing

 

[andyjm]

Lossless refers to the digital data, what you put onto a CD, you get back off again with no loss of the data.

Lossless has no meaning when referring to an analogue system (or the digital sampling of an analogue system). There is always disortion, noise and bandwidth limitation to a greater or lesser extent.

 

[the record spot]

Has the OP been on the Hoffman forum where this precise question was asked recently....?

 

[proffski]

I've been reading about streaming and the different file formats recently. I keep reading that CDs are lossless. Could someone explain this to me please? If CDs are lossless at 16/44.1, why do higher res recordings exist? I dont want to get into the discussion of whether you can hear the difference of the high res formats over CD. However, I would like to understand why if the 24 bit recordings contain more information, how can the CD be considered lossless if by comparison data is missing? - I hope that makes sense. Thanks.

This should set the cat amongs the pigeons. Watch out for punctuation, grammar and American spelling though!

Link: http://people.xiph.org/~xiphmont/demo/neil-young.html

 

[MajorFubar]

In terms of digital sampling, ultimately there is no such thing as lossless, but there comes a point where you've sampled so much data that the audio can be reconstructed without perceptible loss. This is the basis of the proven Nyquist-Shannon theorem, which isn't just some wild idea that may or may not be true, but a proven scientific theorem.

 

[andyjm]

... but there comes a point where you've sampled so much data that the audio can be reconstructed without perceptible loss. This is the basis of the proven Nyquist-Shannon theorem, which isn't just some wild idea that may or may not be true, but a proven scientific theorem.

Well no, actually. Nyquist Shannon is about sampling frequency, not bit depth, nor perceptible loss:

If a function x(t) contains no frequencies higher than B hertz, it is completely determined by giving its ordinates at a series of points spaced 1/(2B) seconds apart.

In other words, sample at twice the maximum frequency, and you've got the lot. The theory assumes the sample process is completely accurate and introduces no uncertainty.

The 'audio can be reconstructed without perceptible loss' bit is thanks to Sony and Philips who set the standard at 16bits, 44.1KHz sample rate.

 

[spiny norman]

Link: http://people.xiph.org/~xiphmont/demo/neil-young.html

Again? Really??

(Pigeons ignore cat with a nonchalant 'Yeah sure: been there, done that' shrug)

Mind you, surely this thread (and the whole lossy/lossless thing) is about file compression, not sampling rates and the number of bits?

 

[Alec]

Link: http://people.xiph.org/~xiphmont/demo/neil-young.html

Again? Really??

(Pigeons ignore cat with a nonchalant 'Yeah sure: been there, done that' shrug)

Mind you, surely this thread (and the whole lossy/lossless thing) is about file compression, not sampling rates and the number of bits?

I think that - along with your description of FLAC - needs a little unpacking. You seem to say that information is still taken away (but only for a little while) with FLAC. Isn't it more accurate to say that the data is placed in a kind of digital container that makes it look like there is less, much like a pack-a-mac?

Lossy/lossless has always been talked about with reference to bits as far as I can see...

 

[MajorFubar]

... but there comes a point where you've sampled so much data that the audio can be reconstructed without perceptible loss. This is the basis of the proven Nyquist-Shannon theorem, which isn't just some wild idea that may or may not be true, but a proven scientific theorem.

Well no, actually. Nyquist Shannon is about sampling frequency, not bit depth, nor perceptible loss:

If a function x(t) contains no frequencies higher than B hertz, it is completely determined by giving its ordinates at a series of points spaced 1/(2B) seconds apart.

In other words, sample at twice the maximum frequency, and you've got the lot. The theory assumes the sample process is completely accurate and introduces no uncertainty.

The 'audio can be reconstructed without perceptible loss' bit is thanks to Sony and Philips who set the standard at 16bits, 44.1KHz sample rate.

i know now all that. I was attempting to explain it in a way the op might follow. But the major thing to take on board is that 16/44 is good enough for a myriad of technical reasons that mere mortals like us (well me definitely) don't need to worry about.

 

[inbox4]

Thanks all that's really helpful, I was wrongfully thinking in terms of the quality of data. I know realise that's irrelevant in terms of lossy/ lossless descriptions, it's just all about whether the encoding used is an exact copy or or a compression hence the two terms.

 

[spiny norman]

You seem to say that information is still taken away (but only for a little while) with FLAC

Don't think I ever suggested that information is taken away, only that it's compressed (ie packed into a smaller space) and then de-compressed

 

[jcbrum]

Just for the record, (pun, groan) a 24/192 file does contain more data (information ?) than a 16/44 file but it is useless, or 'empty' information which cannot be used to create any useful sonic information for human hearing.

Bats may be interested in a frequency response extending to over 90 kHz, but humans stop at c.20kHz. Similarly a dynamic range of over 90dB is wasted in any real world ambient listening environment for human's hearing.

For example, the best vinyl or analogue magnetic tape system, can only utilise an equivalent 12 bits of information (or thereabouts). Many think that is as good as it can get.

JC

 

[Jota180]

I've been reading about streaming and the different file formats recently. I keep reading that CDs are lossless. Could someone explain this to me please? If CDs are lossless at 16/44.1, why do higher res recordings exist? I dont want to get into the discussion of whether you can hear the difference of the high res formats over CD. However, I would like to understand why if the 24 bit recordings contain more information, how can the CD be considered lossless if by comparison data is missing? - I hope that makes sense. Thanks.

This should set the cat amongs the pigeons. Watch out for punctuation, grammar and American spelling though!

Link: http://people.xiph.org/~xiphmont/demo/neil-young.html

Loved the analogy with the eyes and laughing at the idea of videophiles banging on about the wonderful infrareds and ultraviolets in their favourite movie.

 

[Overdose]

Higher than CD resolution recordings are always called 'high resolution' and are nearly always more expensive than the CD of the orignal master.

Why are they more expensive? I'll let you make your mind up on that one, but the original recording process is always at 24 bit or higher anyway, before the downmix to 16 bit for CD.

Some may say it's a cynical marketing ploy to separate the more susceptible from their cash, particularly when you bear in mind that CD resolution did not happen by chance and is derived from the Nyquist theorem to fully encapsulate the audible frequency spectrum.