Talk:Red Book (audio CD standard)

From Wikipedia, the free encyclopedia

This article may be too technical for a general audience.
Please help improve this article by providing more context and better explanations of technical details to make it more accessible, without removing technical details.

Contents

[edit] Range of values

From the article, prior to my edit: "Sample values range from -32768 to +32767, which map from -1.0 to +1.0 when converting to float. (Is this correct? It may be that the lower bound is -32767)". The question at the end belongs in here, rather than out there. So now it is. --203.206.183.160 13:05, 20 July 2006 (UTC)

Don't know the answer myself, but this might help: It depends on the format in which the integer values are stored. A two's complement representation will range from -32768 to +32767, whereas a signed integer representation will have a lower bound of -32767 (the total number of integers is the same in any 16-bit representation, but signed integer has two zeros (+0 and -0), so two's complement allows one more number. --203.206.183.160 13:05, 20 July 2006 (UTC)

from dave:
Right that's my point exactly. What *is* the red book format stored on a standard CDDA disc? It makes life much easier if it really is -32767..+32767, that way you just divide by 32767 to get your float value. Else you have to do this:

valF = <integer val>;
if (valF == 0) {
  valF = 0;
} else if (valF > 0) {
  valF /= 32767.0f;
} else {
  valF /= 32768.0f;
}

Which is cumbersome. Also you can't perfectly cancel an integer signal (via subtracting it's negative) due to aliasing (since the ranges for positive and negative are actually different). There must be some red-book aware guru who can settle this matter? --davec 21:58, 31 July 2006 (UTC)

I don't understand what the relevance of interpreting the numbers as floating point actually is? The integer number representation represents a digital code value that is fed to a digital to analogue converter, which then converts the digital integer value into an output voltage or current. D-A converters do not - not that I have ever seen - take as input any kind of floating point numbers. They use either signed or unsigned integer numbers.

The only reason you might want to convert the integer form to floating point is if you wish to do some further processing such as the application of digital filtering algorithms.

And if you really do need to represent the sample value in floating point form, then you're only losing out one code value in the range of 65536 possible codes, and it's a maximum level value of 0dBFS, maximum volume level which hardly anybody mixes/masters the music at that level because it will end up being dangerously loud, so the small error in format conversion from integer to floating point is of little concern.

I'm not asking for your opinion or your interpretation of it's relevance. I'm asking *what is the spec*. If you know, just say what it is, if you don't know, well, then you don't know. So I ask again, What is the spec? davec 23:11, 20 June 2007 (UTC)
Look, just do the math. A sample has a depth of 16 bits. Therefore, there are 216 = 65,536 possible levels of amplitude which can be represented. This makes it clear that the range cannot be only -32,727 to +32,767. Since the integer is two's complement signed, it represents values from − 2(16 − 1) to 2(16 − 1) − 1. Therefore, the range is -32,768 to +32,767. tgies 06:01, 22 June 2007 (UTC)

I don't know what this floating point stuff is about, but the data portion (the actual samples, 6 per frame) is PCM as in a WAV file and is (correctly stated here) as ranging from -32768 to 32767 (the sampleing order is interleaved but the samples are the same as WAV). jrg 17 July.

[edit] Mastering formats

The Sony PAL format U-Matic 3/4" videotape was the original media for mastering Redbook Audio. The 44.1Khz sampling rate is directly related to this choice of media, not to any considerations about Nyquist Frequency or any of the other reasons given for why 48Khz was not chosen, which was the sample rate used for most other digital audio in the then new field.

http://www.edinformatics.com/inventions_inventors/compact_disc.htm

A Sony 1610 or 1630 PCM digital audio converter for U-Matic tape can store 6 samples (3 for each stereo channel) per video line.

PAL video (as recorded by a U-Matic VTR) has 294 lines and 50 fields per second, which gives 44,100 samples/second.

NTSC can store more samples per second due to its higher frame/field rate, but its higher capacity wasn't used. The most likely reasons being that more countries use the PAL TV standard than the NTSC standard, which would make PAL U-Matic VTR machines more common, and that the higher rate attainable would allow fewer minutes of music to be stored on an Audio CD.

NTSC recorders were not used - 60Hz monochrome units were. While 60Hz video has more frames/second, it has fewer lines/frame. According to John Watkinson in "an introduction to Digital Audio" (section 2.7) both 60Hz and 50Hz recorders give the same 44.1kHz figure. 60Hz video has 35 blanked lines while 50Hz video has 37. At 3 samples per video line, 3 x 60 x (525-35) / 2 = 3 x 50 x (625-37) / 2 = 44100 TheGiantHogweed 08:22, 18 July 2007 (UTC)

The 1610/1630 U-Matic (or Umatic) audio CD mastering format has held on until the past 2 to 4 years as the standard that CD mastering companies accepted above all others. As the U-Matic VTRs have gone out of service due to lack of replacement parts, other digital tape formats have become more popular for mastering. Yet even those are becoming less used due to the improvements in CD recorders "burn" quality at low speeds and the improvements in error checking at the mastering stage.

[edit] too technical

I'm adding the {{technical}} tag to this page. The meat of the article, describing the nature of the spec, needs some expansion I think. Perhaps a little about how the spec is used by manufacturers. I'm a relatively technical person and I was completely bewildered by the middle passage.Jasongetsdown 22:31, 15 November 2005 (UTC)

[edit] merge

This article should be merged to Rainbow Books. Please comment at Talk:Rainbow Books. —Quarl (talk) 2006-01-15 11:42Z

I think this is a bad idea. The Rainbow Books page has its own content, and serves as a nice index to the various Book pages (Red Book, Yellow Book, etc.). There is no reason to merge it in here, and doing so would undoubtedly cause a loss of information (contextual or otherwise) in the process. I am strongly against. --Kadin2048 18:10, 23 May 2006 (UTC)
Copied comment (above) to Talk:Rainbow Books where discussion is held --H2g2bob 15:57, 22 June 2006 (UTC)
rm. tag --H2g2bob 21:45, 25 June 2006 (UTC)
I agree with Kadin2048 - I did not know of the term 'Rainbow books', only of the 'red book' spec (and of the various books describing the postscript spec, but that's not related to audio, although it may be included in the 'Rainbow books' series..). At best, a redirect (preferably directly to a section, if that is possible with WikiMeda) from 'Red Book' to a section of the 'Rainbow Books' article should be maintained.

[edit] De-facto vs. specified

"It also specifies the form of digital audio encoding (2-channel 16-bit PCM clocked at 44100 Hz). These parameters have become something of a de-facto standard."

How can a standard be both specified and de-facto? Tgies 06:07, 20 February 2006 (UTC)

I think what the author was trying to say may have been something like:
"It (the Red book standard) also specifies the form of digital audio encoding for audio CDs (2 channel stereo 16-bit PCM encoding at a 44.1 kHz sampling rate). These parameters have become something of a de-facto standard for all digital audio." T3h933k 01:30, 22 March 2006 (UTC)

[edit] 16 bits per channel or 16 bits total?

This article implies that CD audio uses 16 bits per channel (in the maths calculating the bit-rate)... However, the bit rate that they come up with would mean that 74 minutes of audio (before error correction codes) would require 783MB to store ( 1411.2 * 60 * 74 / 8 = 783216KB ). If it were 16 bits for the two channels, this would mean it would require only 392MB before error correction codes, which seems more realistic given that a 74 minute CD holds 650MB of data. Can anyone more knowledgable comment on this? Thanks. -- David Scarlett(Talk) 07:01, 28 April 2006 (UTC)

Nevermind, apparently it is 16 bits per channel and you can fit more audio data than general data due to different error correction schemes, which would explain why the audio data can take up more than the generic data capacity of the disc. This article explains it. -- David Scarlett(Talk) 12:31, 28 April 2006 (UTC)

[edit] "Text" logo

Image:Cdat.jpg

Can someone add to the article what the "Text" caption means in the Compact Disc Digital Audio logo? (see right.) Tempshill 01:50, 19 June 2006 (UTC)

Maybe it is "CD-Text" as describe at http://www.true-audio.com/Compact_disc --Joshtek 14:32, 20 July 2006 (UTC)

[edit] Standard?

The Red Book is the informal name for the Compact Disc Digital Audio specification (more commonly known as CD-Audio). While the word "standard" is quite similar, it doesn't quite mean the same thing. Products such as CDs are produced to meet a specification. When they are tested, they either meet the specification or they don't. Specifications are specific. Standards can be something less specific (de-facto, etc.). I would suggest that this article be renamed Red Book (CD-Audio specification). Tvaughan1 13:36, 28 August 2006 (UTC)

[edit] Removed obsolescence section

I removed the section arguing that CDs are obsolete due to the popularity of iTunes/ iPod. This section cited no sources and seemed irrelevant to the topic of the blue book standard.

[edit] Licensing confusion

Is it really $5000 to license the standard itself? I am guessing it's $5000 for a manufacturer of audio CDs to implement the standard. Anyone can get the publication for $200. —mjb 23:01, 21 February 2007 (UTC)

[edit] Frequency response?

Does Red book indeed specify frequency response? It seems rather illogical.

The sampling frequency of 44 kHz implies a maximum sound frequency of 22 kHz. I have no idea where the lower limit comes from, but a lower could have been imposed to avoid damage to the user's equipment (Turritus 12:26, 24 April 2007 (UTC)).

22kHz is a theoretical upper bound for the frequency content, but practically, not acheived. The actual boundaries of the frequency response depend on the CD player and could be quite difficult to compute. Perfect reconstruction up to 22kHz would require the CD player to compute a large series of Sinc functions for every sample. See Nyquist–Shannon sampling theorem for a full explanation. Sirhans 22:19, 2 May 2007 (UTC)

The 20Hz lower limit is likely not a technical limitation of the red book standard, but was simply added because it's the generally-accepted (and often-cited) lower limit of human hearing -- the point below which supposedly 50% of people cannot hear the sound.

I certainly think frequency response is an incorrect choice of words for the concept... an electrical system has a frequency response, a set of data does not. A CD is a collection of data and therefore the latter.

By that analogy, so is a vinyl record or cassette. Irrespective if they contain data or not, they become part of an electrical system when played back. Therefore, frequency response of the CD format is absolutely a correct choice of words. Vinyl does not go down to 20hz, for example. —Preceding unsigned comment added by 71.62.11.217 (talk) 23:53, 29 September 2007 (UTC)

[edit] Kbps

Just a little note: According to our article on Kbps (data unit, following the disambiguation), a Kbps = 1000 Kbps. This is the standard typically used for digital audio and storage, so I think we should follow it. I recalculated and changed the nominal bitrate in the article from Kbps == 1024 bps to Kbps = 1000 bps. 4.242.108.211 (talk) 20:03, 25 March 2008 (UTC)

I notice the article likes to often use different units: kbits here, KiBs there. Please unify. Jidanni (talk) 00:52, 20 May 2008 (UTC)
The different units used are appropriate in context. tgies (talk) 09:33, 20 May 2008 (UTC)