Talk:Comparison of audio coding formats

	This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.ComputingWikipedia:WikiProject ComputingTemplate:WikiProject ComputingComputing
???	This article has not yet received a rating on Wikipedia's content assessment scale.
???	This article has not yet received a rating on the project's importance scale.
	This article is supported by WikiProject Software.

Suggestions

This article needs expanding and updating. Ace ofgabriel (talk) 18:30, 22 March 2008 (UTC) The licenses section should be limited to three or so of the more popular tools (e.g.: envolve encoding, for the most part).[reply]

Latest stable version could use some work. Perhaps split into reference tool version, and spec. version?

Perhaps narrow down the *nix OSs to just "*nix". I am aware that there is an open source decoder in the works for ALAC, however I think it's not released yet.

mplayer 1.0pre7 and above has native support for decoding ALAC.

The stereo column could use some detail as to what mid/side, intensity, etc mean. Also do we need a mono column? I do not think bits per sample applies to lossy codecs (for the most part store data in frequency domain, and most are able to output into floating point). Perhaps some explaination is due as well: bits per sample here means of the original waveform, not of the stored compressed version (see Wavpack's rate description). For the values in parenthesis I mean that the format specification supports the values, but the software available doesn't fully support it.

Some links to debates about which format is best would improve the article greatly. Cacophony 21:26, 29 December 2005 (UTC)[reply]

What is 'Hybrid' in algorithm in technical details table? would've been better if there was more info on that.. 61.17.227.44 11:30, 9 March 2007 (UTC)[reply]

Probably a mistake on the sample rates

I'm not sure, but are all the sample rates correct? WMA, for example, has 48000kHz. Isn't it 48kHz? Some other values are strange too. --200.165.128.150 14:42, 28 February 2006 (UTC)[reply]

Cleanup needed

This article needs work. What is its encyclopedic context? Is it a sub-page of audio codec? If so, it should have a {{main}} tag.

What is the significance of the audio codecs selected? Do they each form a large, well-documented share of audio files? If not, why were countless other codecs excluded from this list? Why is WMA stated to be unsupported on Mac OS X, when a Windows Media Player has been available on the platform and Flip4Mac provides current support?

Why is the latest stable version mentioned? How are the stable version numbers being verified? Whose implementation is considered to be "the real codec" and why? —donhalcon^╤ 20:10, 7 March 2006 (UTC)[reply]

None of the Category:Software comparison articles are really sub-pages. They have been treated as encyclopedic as stand-alone entities. However, this aricle might benefit from the expansion of the short introductory paragraph (sofixit).

Note there is a request for expansion, so I won't comment on included/excluded codecs.

I do agree that WMA should probably state "supported" on OS X.

The latest stable version provides a check on how current the comparison is. These numbers may easily be verified by any number of methods (not the least of which is to check back with the documentation for the codec, itself. The version numbers and features appear to be for the reference implementation, but the OS support seems to be ANY implementation. Perhaps this table can be expanded to say whether there is a reference implementation or whether the only implementations are by someone other than the original author (particularly on the *nixes). --Karnesky 20:08, 7 March 2006 (UTC)[reply]

This Article needs expansion and updating. Ace ofgabriel (talk) 18:29, 22 March 2008 (UTC)[reply]

Reliable sources

What do people here consider as reliable sources? My company has developed an audio codec which is mentioned in an article I am working on here, and would like to know where editors working on this comparison expect information to be published. Stephen B Streater 20:31, 9 June 2006 (UTC)[reply]

More codecs that should be added.

Of course, there's more at hydrogenaudio, but there's a couple more.

Some issues

For the purposes of this article, what is the difference between "BSD" and "Unix" (from the "Operating system support" section)?
The statement "this enables their use on any POSIX-complaint system" (from the same section, footnotes) is misleading. Typically, a string of other libraries is required (which are not part of POSIX), and some type of output system (ALSA, OSS, ...; also not part of POSIX).
The use of "OSI" to refer to the open-source implementations is a bit confusing. Simply because the OSI approved the licenses under which the software was published does not mean that they endorse that piece of software, nor that that piece of software has any other affiliation with OSI. Perhaps "OSS" (open source software) or just "Open-source" would be more appropriate.
Perhaps there should be some mention of whether samples are floating-point or integer in the heading "bits per sample" in the "Technical details" section.
How exactly does WavPack do "2.2" bits per sample in lossy mode? Does every fifth sample get an extra bit?
Is a "Monaural" column really necessary? Can it not be assumed that all audio formats support one channel of audio?

—Kbolino 10:03, 17 July 2006 (UTC)[reply]

Implication of inferiority of patented formats

By using red for patented formats and green for free formats, we are clearly implying their inferiority, which in my neck of the woods is called POV. Recommend neutral coloring for that section. - 211.28.81.162 03:24, 20 November 2006 (UTC)[reply]

Agreed. Since this issue has been around for a while & there have been no arguments FOR the use of the "but no" and "but yes" templates, I'm going to remove them. --Karnesky 01:46, 13 March 2007 (UTC)[reply]

But it's still non-neutral POV! The patented ones are now green and the free ones are red, again, implying inferiority of the free ones. This is no good either, this is not the same as a feature checklist where it makes sense to imply missing features are somehow inferior. Not being encumbered by patents is not a missing feature. The neutral thing to do in this case is to just use the words "yes" or "no" without the colorizing templates; or even better perhaps the descriptive more neutral words "patented" and "unencumbered". Use of colors in this column adds no value at all and just implies favoritism one way or the other. - Dmeranda 21:46, 5 April 2007 (UTC)[reply]

"green" means "yes" and "red" means "no." Neither means good/bad/inferior/superior/etc. --Karnesky 22:00, 5 April 2007 (UTC)[reply]

For more than one person, they have meant good/bad/inferior/superior/etc, personally it was that way, even before reading what it says in the box, it seems that what is unencumbered is "evil". Colors should be removed. --Nando.sm 06:51, 12 April 2007 (UTC)[reply]

Further in technical details, it still seems "bad" that not having a constant bit rate. What could be done, and if you like (i know this is for everyone, and not for only a small group of people, but if they agree) to have checks and crosses for has and has not. Don't know if it could work, but options have to be given. --Nando.sm

On the template talk pages, there was general consensus that green meant yes, rather than "good." It seems inappropriate to disregard this recent opinion which was advertised on all pages which used the templates for one particular page with far fewer readers. There was also a poll on checks, etc. If you think the templates should be improved, clearly the place to discuss it is on the template talk page. That the templates are called and have the self-proclaimed mission of substituting for "yes" and "no" suggests they can be appropriate in those columns. --Karnesky 16:07, 12 April 2007 (UTC)[reply]

It seemed evil before, but now it looks ugly... i will be killed for this, but wikipedia has reached concensus, so I will accept it. For information about color tags in tables go to table cell templates. Now in other things why not use the yes2 and no2 they are not so color saturated so it could have less appeal to the human eyes. Nando.sm 17:35, 30 May 2007 (UTC)[reply]

The question is "Patented" the answer is "yes" so it should be green, but it is red. Similarly with DRM. Also, free is green, with cost is red, and OSI implementations are green, proprietary implementations are red. (I see it's actually magenta and cyan, not red and green, but give me a break!) The table is ugly and therefore hard to read, and still expresses bias. The switch to magenta and cyan just dodges the issue. I suggest shades of tan/yellow/off-white/etc. 216.254.64.90 (talk) 19:43, 31 December 2008 (UTC)[reply]

What constitutes "one codec"?

Some of the format compared in this article really comprise multiple separate codecs, in that no decoder program could decode all variants of them. For example, "AAC" lumps together several variants: LC-AAC (the standard version), HE-AAC (an enhanced version), HE-AAC v2 (a revision of HE-AAC), etc. A decoder that only understands the baseline LC-AAC won't be able to play back HE-AAC. By comparison, any reference-compliant MP3 decoder should be able to decode any MP3 file... whether it's encoded with some ancient Fraunhofer encoder, or the latest version of LAME.

In addition to AAC consisting of multiple incompatible codecs, ATRAC also has incompatible variants, and RealAudio uses numerous variant codecs (it's more of a container format than a codec). WMA may refer to multiple incompatible codecs as well (WMA 1/2, WMA Pro, WMA lossless, WMA voice). It seems that these "codecs" should be more properly described as brand names of loosely related codecs.

Anyone have any suggestions as to how to convey this confusing state of affairs in the article? MOXFYRE (contrib) 17:53, 5 October 2007 (UTC)[reply]

Comparing codecs or formats

I find it very unclear if we're comparing codecs or formats here. The title says “Comparison of audio codecs” while the table apparently compares formats — i.e. AAC to MP3 instead of LAME MP3 codec against Nero MP3 and QuickTime AAC. The table even has an “implementation” entry. Shouldn't the comparison be between those? —Preceding unsigned comment added by 81.243.206.5 (talk) 10:43, 4 March 2008 (UTC)[reply]

I agree that this article is not correct. Audio formats are not audio codecs. RealAudio is not a codec, it's multiple codecs that constitute a format. Some of the information here needs to be moved and/or deleted. Sapwood2 18:32, 13 April 2008 (UTC)[reply]

ffdshow & ProRes 422

Where would ffdshow & ProRes 422 fall in this comparison? —IncidentFlux ^{[ TalkBack | Contributions ]} 17:37, 31 October 2008 (UTC)[reply]

FP means...

Several references to "fp internally" are used in one table without any reference to what "fp" means. Could someone in the know please clear this up? User:Michael Daly --24.138.146.91 (talk) 02:39, 27 December 2008 (UTC)[reply]

It means floating point. Although it's a little misleading: some formats (like MP3) that are usually implemented with floating point have fairly tight integer constraints internally and as a result don't have (much) more dynamic range than 16bit audio. --96.231.29.35 (talk) 17:08, 31 May 2009 (UTC)[reply]

Avoiding a revert war

A AIP keeps undoing the change I made that MPEG-1 Layer II audio is not under any active patents. All current MPEG-1 related articles, with the exception of this one, reports no patents are being enforced against implementations of MPEG-1 Layer II. Given the MPEG and ITU processes that encourage patent exposure, it's less likely that any active patents apply to this pre-1991 technology than, say, Ogg Vorbis.

As an aside, reverting my change demanding "I explain first", when I gave full explanations in the summary field, and when you've made no attempt to explain your position, seems like a very poor breach of etiquette.

I think the way to go here is to work on the baseline that if someone cannot cite a patent, the page should not claim a format is patented. If there are no serious objections, I'll go forward tomorrow and update the Layer II information. --208.152.231.254 (talk) 14:23, 4 September 2009 (UTC)[reply]

It's not "a" AIP, but two (and not the same person), so I would suggest to wait more than 24 hours. Are you really convinced that none of the (several) patents that apply to MP3 do not apply to MP2? --CE 62.178.80.242 (talk) 16:18, 4 September 2009 (UTC)[reply]

You're correct, I thought it was one.

However: First, we don't need a question answered with another question, still less one that expects me to prove a negative, but an answer. If MPEG 1 Layer II audio has a valid patent being enforced against it, then you should be able to cite it. To put it another way, you should be able to stick <ref>US Patent 391837491</ref> after the claim that it's patented. You can't expect me to prove a negative. Would you really feel that it's valid to put <ref>It hasn't been proven that the patents applying to MP3, which deviates from MP2 in many substantial and dramatic ways, do not apply to MP2, which is a format based on technologies from the mid-eighties</ref>

As I said above "Given the MPEG and ITU processes that encourage patent exposure, it's less likely that any active patents apply to this pre-1991 technology than, say, Ogg Vorbis."

If nobody here can list a specific patent applying to MPEG 1 Layer II audio, then we need either to mark the patent status of MPEG 1 Layer II and all formats currently marked as free (including Ogg Vorbis) as "Unknown", or we need to mark it as free. It's not hard. It's a simple question: if patents are being enforced against MPEG 1 Layer II, you should be able to find a patent that's being enforced. 208.152.231.254 (talk) 17:52, 4 September 2009 (UTC)[reply]

I added now a reference about some patent issues. - Kuro5hin.org (2008-07-20) Patent Status of MPEG-1,H.261 and MPEG-2. I would like to notice that "MP2" (MPEG Layer II) is also defined in MPEG-2 Part 3. Maybe there are different patents for MPEG-1 and MPEG-2 audio. --89.173.68.106 (talk) 20:02, 5 September 2009 (UTC)[reply]

The above reference (with others) is used by the MPEG-1 article to claim that MPEG-1 is patent free with the exception of MPEG 1 Layer III Audio. I do believe that Fraunhoffer claims patents with the surround sound extensions to MPEG-2 Layer 1-3 audio. I wonder if we should separate MPEG 1 and MPEG 2 out here, as the patent situation is clearly different with both. 208.152.231.254 (talk) 14:00, 14 September 2009 (UTC)[reply]

FYI, RedHat specifically refuses to include TwoLame, a MPEG-1 Audio Layer II encoder because of patent issues [1] On the whatwg mailing list George Maxwell listed US 5,214,678 [2] as a possible Layer II patent. Jrincayc (talk) 03:21, 6 October 2009 (UTC)[reply]

Split lossy and lossless formats

This article is becoming more messy by the hour, I would suggest a split of lossless and lossy formats, because it will reduce the clutter and because the applications of lossless (archival) and lossy (bandwidth and space savings) are quite different. Maybe even split formats for voice transmission and make this article "comparison of music formats" (rather arbitrary :( ) or something the like.

Kohlrabi (talk) 12:59, 18 December 2009 (UTC)[reply]

I think splitting can be complicated. There are some hybrid compression formats, that can be lossy and also lossless - e.g. MPEG-4 SLS, WavPack, OptimFROG. Splitting voice transmission and music formats is also not very easy, because there are attempts to create universal formats - like CELT, MPEG-4 Part 3 formats (like USAC - Unified Speech and Audio Coding, AAC-LD and others), G.719, Siren 22, AMR-WB+, HE-AAC v2 ... etc.

--89.173.68.106 (talk) 20:17, 19 December 2009 (UTC)[reply]