Talk:MP3

The intro is a mess

Ok so we all know that everyone doesnt know about mp3's but if you dont know then u shouldnt be editing this page!!! all this 'i think' and 'im not sure' in the intro makes it look horrible. can someone remove this or verify the information to fix this? 75.105.14.238 (talk) 16:21, 30 June 2008 (UTC)

I agree. This text is poorly written, plus it delves too much into technical detail for the introductory section of this article. Furthermore, the topic of variable bit rates is already discussed later on in the article. The new text has no references, and even explicitly doubts itself, so it is worse than unverifiable. I am going to remove it. If the contributor wishes to rework portions of the text as cited and verifiable information (i.e. no "I'm not sure" or "maybe" stuff) to the existing section about variable bit rates, that would be great. CosineKitty (talk) 17:48, 30 June 2008 (UTC)

Magic Number

According to the article the "magic number" of MP3 is "ID3". It's not actually, what shows is the magic number of the ID3v1 tag that is present at the beginning of lots of MP3 files. MP3 files without ID3 tag, or with an ID3v2 tag (which is at the end of the file) won't show "ID3" in the first 3 bytes. As far as I know there is no magic number for MP3 files. —Preceding unsigned comment added by Jaho (talk • contribs) 22:15, 17 March 2008 (UTC)

Sorry, forgot to sign that. Jaho (talk) 22:18, 17 March 2008 (UTC)

You're right, I removed this from the article. --Gabriel Bouvigne (talk) 22:56, 17 March 2008 (UTC)

Bit-rate

The russian music download site allofmp3.com seems to be offering mp3's at a maximum bit-rate of 384kbps using the LAME codec. A qucik google search didn't provide any details on this, wondering if anyone knew anything worth adding.

I think the maximum for LAME mp3 is 320kbps. It is ~500kbps for Ogg's, though. --Russoc4 14:53, 29 June 2006 (UTC)

Update: If they are 384kbps, then they aren't mp3s. Most likely mp2s. From the article:

   * Layer 1: excellent at 384 kbit/s
   * Layer 2: excellent at 256...384 kbit/s, very good at 224...256 kbit/s, good at 192...224 kbit/s
   * Layer 3: excellent at 224...320 kbit/s, very good at 192...224 kbit/s, good at 128...192 kbit/s

LAME free-format allows for up to 640kbps, though practically nothing can decode it. 70.45.49.36 03:57, 11 July 2006 (UTC)

The bit rate of CD-audio is listed as 1411.2 kbps. This is technically incorrect, as there are 1024 bits/kbits, so it's really 16bits * 44100 Hz * 2 ch / 1024 = 1378.125 71.106.82.134 18:15, 6 August 2007 (UTC)Kieran Coghlan

Check out http://en.wikipedia.org/wiki/Kibibit http://en.wikipedia.org/wiki/Byte and http://en.wikipedia.org/wiki/Bit they cover the confusion between using 1024 and 1000 in kb vs. and kibit. —Preceding unsigned comment added by Ozzy 98 (talk • contribs) 20:36, 20 October 2007 (UTC)

The usual 128 and 192 kbps rates, do they account for both channels in stereo? Or maybe it is, say, 128 kbps per channel... you may want to make this issue clearer. Thanks! --Biscay (talk) 15:30, 18 February 2008 (UTC)

Bog standard 128 kbps mp3s can be 64 kbps per channel, though there are differences between modes such as 'joint stereo' and 'high frequency'. In joint stereo mode both channels share some of the bitstream, making it difficult to say exactly how much of 128 is devoted to each channel. True mono 128 kbps mp3s use the whole 128 bits for a single channel and are the same encoding quality as 256 kbps stereo mp3s... except that there are supposedly mono mp3s out there which are actually two channels of identical (or nearly identical) 64 kbps audio. This kind of parallel mono would be half the quality of true mono and the same quality as a stereo mp3 of the same bitrate. Hmmm. Binksternet (talk) 21:16, 19 February 2008 (UTC)

MPEG-I/II

In the first line it says "MPEG-1/2", and this needs to be explained on the page, I think. Here's what I know about it:

Phase 1 can handle input streams (or WAV files) with a sample frequency of 48000, 44100 or 32000 Hz and is therefore used most often, obviously.

Phase 2 will only support stream for 24000, 22050 and 16000 Hz. Basically, Phase 2 is intended for lower bit rates (e.g. for voice communication, or if you need small files with reduced quality, podcasts and live online audio-feeds and the likes).

The lowest bit-rate for Phase 2 is 8 kBits/sec while for Phase 1 the lowest bit rate is 32 kBits/s. 195.64.95.116 23:11, 11 Mar 2005 (UTC)

Response to 195.64.95.116: There are several problems in the above comments. Please refer to an authoritative source such as The official MPEG site or The MPEG Industry Forum site or perhaps the somewhat less official MPEG.ORG site. Be very careful to only get your information from such reliable sources, as there is a significant amount of confusion found on more random web sites. Do not call the MPEG-X numerical suffixes "Phases". Do not use roman numerals to denote them. MP3 originated from MPEG-1 Audio Layer 3, also properly referred to as MPEG-1 Part 3 Layer 3 (where "Part 3" refers to audio coding, "Part 1" is multiplexing, "Part 2" is video coding, etc.). The stuff above about sample frequency looks wrong too. I'm not personally aware of any connection between MP3 and MPEG-2, except for the former being a predecessor of the latter. Pangolin 06:16, 12 Mar 2005 (UTC)

Actually, it is exactly the other way around; You have to call the old MPEG-X Phases, since they are, and always have been. .MP2 and .MP3 used to both even be called .MPA and this was way before the Part 1 or Layer 3 issues came into play. Hey, I know, because I was there testing and using it back in the day. I don't really care what you think it should be, I know what the Phases meant, why they are there in the format, and when they started putting it in. The "Part 1" came into play because of incompatibility with Part 2, which is different from Phase 1 and Phase 2. MP3 originated from MPA, which in turn changed to MPEG Phase 1 Layer 3. To help make it clear, it was decided by its creators (hey, ask them) to use the Roman for the Layers. Furthermore, I don't know what fool put in the MPEG-2 part about 'the new' MP3, but that's a silly thing to do; That new MPEG-2 is not the same as MP2, the new MPEG-2 is no longer describing the Layer or Phase, it isn't even 'downwards' compatible with MP3.195.64.95.116 02:25, 28 Apr 2005 (UTC)

MPEG-2 Audio Layer 3 is also valid, and not limited to low bitrates (i have often seen MPEG-2 transport streams with 128kb/s Layer 3 audio). Furthermore, 8 kb/s is not ISO MPEG-2; it is part of a Fraunhofer sub-spec, known as MPEG-2.5—which is not endorsed by the ISO. MPEG-4 can also contain Layer 3,2,1 streams, as well as VQ and AAC. —Brian Patrie 05:27, 1 December 2005 (UTC)

Why have all references to MPEG-2 and MPEG-2.5 been removed from the first paragraph? As I understand it, MP3 covers MPEG-1 Layer 3, MPEG-2 Layer 3 and MPEG-2.5 Layer 3. At least there are many .mp3 files around that are actually MPEG-2 or MPEG-2.5, and no decoder complains.--87.162.39.233 14:46, 24 June 2007 (UTC)

Piracy

I think there should be a heading regarding the allegations of "piracy" and the RIAA lawsuits. perhaps mention of the mp3/warez "trading" scene? Does anyone agree? Alkivar 04:56, 19 Oct 2004 (UTC)

VERY much agreed. User:Afolkman 1:41, 16 Nov 2004

It would be more appropriate to link to an external article, as this subject is not confined to the MP3 domain. I see, though, that we now have an MP2 and MP3 and the Internet section, so i've linked the word "piracy" to Pirate (disambiguation). I considered linking to copyright infringement, but decided that this would tend to reinforce the media-nurtured impression that the two are synonymous. I added the latter to See also. —Brian Patrie 06:07, 1 December 2005 (UTC)

Summary and Psychoacoustics

There are two things that I don't like about this page:

• I think the one-line summary on the very top of the page should give more information -- one has to trawl way down the page before psychoacoustics are mentioned
• The History section contains useful information, but I think there's too much babble about claimed or presumably more correct bitrates. That part should move to the Quality section.

I've changed the link to psychoacoustics from Pycho-acoustic coding to Psychoacoustics as there is a redirect. --Cpk 21:20, 4 Sep 2004 (UTC)

Legality and Acceptance

From the 2nd to last paragraph: "Of course, until there is widespread acceptance, nobody will bother with litigation and without a clear-cut status, there is unlikely to be wide-spread acceptance."

I disagree with this bit entirely. Smoking marijuana and downloading copyrighted mp3s both have a clearcut legal status--that is, illegal--in many countries but both activities enjoy "wide-spread acceptance." So "of course" the sentence oversteps the mark quite a bit. Furthermore, the sentence preceding it: "I know of no rigorous listening tests to back up the quality claims and the IP questions have not been litigated, so nobody really knows for show what the status is." is from a point-of-view which the wikipedia does not maintain, that is, first-person, and makes another inaccurate claim, which is that no one knows the legal status--activities are legal until litigated otherwise. I'm moving the two sentences here.

(redacted from the paragraph beginning "The Vorbis format") : "I know of no rigorous listening tests to back up the quality claims and the IP questions have not been litigated, so nobody really knows for show what the status is. Of course, until there is widespread acceptance, nobody will bother with litigation and without a clear-cut status, there is unlikely to be wide-spread acceptance." --KQ

Numbers, Parts and Layers

MP3 refers to MPEG-1 Layer 3. MP2 (audio files) refer to MPEG-2 Layer 3. AFAIK, MPEG-2 Layer 3 is basically the same as MPEG-1 Layer 3, with some slightly different packetization. Is it worth even putting it in the list of similar formats? AAC is also known as MPEG-2 AAC, but this probably isn't worth worrying about. -D

MP3 actually refers to all MPEG layer 3 audio. MP3 at sampling frequency at least 32 KHz is called MPEG-1 layer 3 and uses MPEG-1 packets; MP3 at sampling frequency up to 24 KHz is called MPEG-2 layer 3 and uses MPEG-2 packets. "MP2" is primarily MPEG-1 layer 2 audio used in classical MPEG applications such as CD-i and Video CD, but you'll often find MP3 files labeled as MP2 to get them through file type filters on web hosting services. Winamp processes all files named *.mp2 and *.mp3 as generic MPEG audio, sending them to its "Nitrane" MPEG audio decoder. See http://www.mpeg.org/MPEG/MPEG-audio-player.html --PP

Argh. Looks like you're right. I'm still confused, though. According to the MPEG specs, MPEG-1 Layer 2 describes video, Layer 3 describes audio. Are there sub-layers to "Layer 3", and is that what we're talking about? I find this whole thing very messy, and it'd be nice to clean it up on the MPEG pages. -D

Response to -D: Don't confuse "Parts" and "Layers". Part 2 is video. Part 3 is audio. In the case of Audio Layer 3, the term "Layer" refers to a lower level of the hierarchy than the term "Part". Layer 3 is something inside of Part 3. Pangolin 06:51, 12 Mar 2005 (UTC)

Codecs and Algorithms

Hmm, I'm a bit confused. 'codec' is said to be the _same_ as an audio compression algorithm. I would think a codec is a specific _implementation_ of an audio compression algorithm. Am i just plain wrong? or? --arcade

Yes, in my opinion, codec (coder + decoder, analagous to modem being short for modulator + demodulator) is something completely different than an algorithm. While an algorithm could be said to be a set of instructions to yield a desired result, a codec is an implementation of both an algorithm and also the reverse of the same algorithm, to aid media creation and conversion tools. In other words, I think they're two completely different things. I'll fix this problem in the description. Even the article page for Codec that this one links to, says that a codec is a device or program processing the data in some way. I.e. not an algorithm, which is only a set of instructions on how to process the data. --Jugalator

It's even more complex a picture than that. In this case, the standard specifies only the decoder, and says very little about the encoder. So the standard does not specify a codec, only part of one. The algorithm used for encoding is not specified. But, properly, the term algorithm is a general term that can also apply to the specified decoding process by itself. Pangolin 06:51, 12 Mar 2005 (UTC)

NPOV complaints

This article reads like it was written by an audiophile. Choice excerpts:

• To many other listeners, 128 kbit/s is unacceptably low quality, which is unfortunate since many commonly-available encoders set this as their default bitrate.
• It is important to know that despite of all the flaws, recent multiformat listening tests (http://www.rjamorim.com/test/multiformat128/results.html) once again show that LAME MP3 easily rivals its technological successor AAC. (Vorbis aoTuV is tied with Musepack at first place, Lame MP3 is tied with iTunes AAC at second place, WMA Standard is in third place and Atrac3 gets last place). (bold original)

Most of the stuff under the encoder comparison is also POVish. The Alternatives is similar. Random speculation has worked its way in. Finally, the Online Music Resources is marginal. - Fennec (はさばくのきつね) 03:34, 10 Jul 2004 (UTC)

What part of that "despite of all the flaws" did you find 'audiophile'-ish? You can't disregard the fact that inherent to this format (MP3) there are quality-issues involved. Everybody wants (and needs) to know that, in order to understand what MP3 is about. So, to then state under some Quality-section that MP3 is some kind of ugly sounding bad quality trashy format, seems very unfair to me. Check these if you are not convinced; http://www.heise.de/ct/00/06/092/ (I believe there is a translated version of it somewhere) http://jthz.com/mp3/#MYTH So, that is why I posted the bold part; use the right encoder, with the right config, and you'll have flawless quality MP3 encoding. It's been proven. 195.64.95.116 01:30, 5 Sep 2004 (UTC)

I agree, this does read like it was written by an audiophile. Feel free to edit if you don't like it.

The entire MP3 quality section reeks of NPOV and unverified information, especially statements like this "However, listening tests show that with a bit of practice many listeners can reliably distinguish 128 kbit/s MP3s from CD originals [...] reaching the point where they consider the MP3 audio to be of unacceptably low quality." Needs a rewrite for sure (and not by some audiophile with no sources). 70.45.49.36 04:06, 11 July 2006 (UTC)

I find the main text perfectly reasonable, and these complaints hard to follow and inconsistent. MP3 is an audio format. I have no clue if the author considers him/herself to be an "audiophile", but certainly I see no shame in being an audiophile, or any problem in the idea that an article about an audio format should be written by an audiophile. Who better ? someone who is not interested in sound and isnt able to distinguish high quality and low quality reproduction ? I personally DO consider myself to be an audiophile, yet am almost ashamed to say that I sometimes struggle to distinguish 128 bit MP3 from a CD original. So its good. But its not perfect and under ideal conditions I can easily tell, and for thius reason I consider 128 bit unacceptable, and always rip at max quality 320 bit VBR. Which I think fits perfectly with the authors text. Whats the problem guys ? / Dave smith

The problem is no sources. Even an audiophile would be OK as a source, if published and referencable, but an editor's own audio opinions are not encyclopedic. Dicklyon 18:57, 25 July 2006 (UTC)

As above. It doesn't matter how plausible the text is or how well it jives with your subjective experience. The article makes several unsourced claims about relative "quality" (which is not defined in specific terms) of MP3s of various bitrates, and it makes several unsourced claims that people who make claims about audio quality have been proven to be unreliable. It's not unreasonable to demand sources be cited or the section be rewritten or deleted.

Not related to the article, but responding to your comments, you should be aware that using VBR results in the lowest possible bitrate being chosen for each frame to maintain (not drop below) a certain level of quality (and in LAME at least, that quality level is verified by trial and error). If you want max quality per frame, you'd be using 320 kbps CBR, so that the encoder is under no pressure to drop the bitrate on frames where it thinks it can get away with it. However, 320 kbps CBR is unreliable in some encoders (even LAME), introducing a kind of ringing artifact in the high end, depending on what other settings you use (highpass filter, psy model) and the content of the audio. —mjb 19:11, 25 July 2006 (UTC)

I'd just like to say that i think an audiophile is the perfect candidate for writing on this sort of topic if it had good sources. Now i know ive come across articles on the internet (not blogs) that have tested mp3 satisfaction on subjects, it just a matter of finding them. ive read the section and it all seems to match up with what i rememeber so i think leaving the [citation needed] will do for now. also, im goin to activly look for said articles whenever i can now. also, i propose that the NPOV tag get removed and replaced with a {{Unreferenced}}. -(chubbstar) ^{— talk | contrib | 15:42, 26 July 2006 (UTC)}

Quote:

• MP3 files encoded with a lower bit rate will generally play back at a lower quality. With too low a bit rate, "compression artifacts" (i.e., sounds that were not present in the original recording) may appear in the reproduction. A good demonstration of compression artifacts is provided by the sound of applause: it is hard to compress because of its randomness and sharp attacks. Therefore compression artifacts are audible as ringing or pre-echo.

Isn't the mp3 compression scheme 'adaptive' in some sense that it only throws out perceptively irrelevant data-subspaces? It seems like ringing/pre-echo artifacts would be associated with other forms of stuff, not medium-quality mp3... is there a source to verify this? Ninjagecko 04:24, 7 August 2006 (UTC)

It does try to make sure the errors due to coding are perceptually small, or masked. But since the filters are essentially time-symmetric, it puts as much error before a transient, where it's likely to be audible, as after, where it's more likely to be masked. That is the scheme is not perfect, and some of its imperfections are these "precursor" sounds. I don't have a reference about who finds these to be audible at what bit rate. Dicklyon 04:41, 7 August 2006 (UTC)

I rewrote the quality section and provided references. I thus removed the NPOV and references tag of the article, as my opinion is that this section is now based and supported by facts.--Gabriel Bouvigne 15:06, 14 September 2006 (UTC)

I must point out that MPEG's own verification tests show that MP3 is substantially worse in encoded quality at a given bit rate than MPEG-2 AAC LC profile.

Woodinville (talk) 23:11, 15 January 2008 (UTC)

MP3: Now with NPOV

The old article was full of awful propaganda, so I revamped it and moved some things into the LAME article. D. G. 10:44, 4 Sep 2004 (UTC)

I kinda disagree with you on that. As with Winamp, also LAME is very much to blame for the grand success of the format. Let's not forget that this was one of the first and highest quality FREE MP3 encoders out there (next to Blade), whilst others needed payment or licensing before being used. So, when speaking of MP3, one needs to speak of LAME. The one would never be this popular without the other. I would hardly call that propaganda, or POV, it's simple fact. 195.64.95.116 01:21, 5 Sep 2004 (UTC)

I'm not sure what you're proposing is a "proven scientific fact." There's no such thing as a proven scientific fact. Anyway, anything is "up for discussion." You simply can't make an edit and declare that it is unquestionable. I believe you are saying that it's a "scientific fact" that nobody can detect the difference between 256kbps MP3 and an uncompressed source. That is untrue, although almost nobody can tell the difference, some golden ears listeners can on certain samples. If you're going to insert these statements you should back them up. You cannot just say "numerous listening tests."

I deleted the paragraph from the design limitations section, because this is already discussed in the MP3 quality section. Please clarify it there if you want. Rhobite 23:01, Oct 3, 2004 (UTC)

It is proven scientific fact that 1 added to 1 makes 2. If you can't get that, you're not worth my time, and I'm not going to discuss proven facts concerning our hearing or MP3 quality, I have better things to do. If you want to state silly 'claims' on MP3 quality which have no base whatsoever other than gossipy (trying to sound like an expert) speak, I'm going against that. I would refer to rec.audio.pro and being a member of the Audio Engineering Society. The point where experts, the high-end listeners and the likes, will not be able to distinguish the MP3 file from the original currently lies around 180 kbps VBR, and 224 kbps CBR mp3 files. This is where 50% of them will say that the mp3 is the original and/or vice versa, i.e. where they can't tell the difference. (This is researched on using LAME.)195.64.95.116 18:53, 4 Oct 2004 (UTC)

Can you link to a study, other than that German study from 4 years ago? Can you link to any specific posts on rec.audio.pro?

You can all test it yourselves, this is quite easy to do. If you can't understand that, you don't belong in this discussion anyway. It's like discussing existence of gravity, or the magnetic polar fields. They are there, and you can't babble on about it the way you want to. Furthermore, the articles in rec.audio.pro or elsewhere would not be read or understood by you anyway. 195.64.95.116 16:04, 31 Dec 2004 (UTC)

The burden is on you. If you refuse to discuss your edits or link to any studies, you are not "worth my time" either. Please do not curse in your edit summaries, and do not personally attack me.

I do as I see fit thank you very much. You must have been deserving of me personally attacking you. 195.64.95.116 16:04, 31 Dec 2004 (UTC)

MP3 and CDDA are not directly comparable. I will not defer to your self-proclaimed "authority."

Of course they are; they are both the end-medium formats people listen to.195.64.95.116 16:04, 31 Dec 2004 (UTC)

Also, if you insist on reverting, please don't reintroduce your own grammar and spelling errors. Rhobite 19:13, Oct 4, 2004 (UTC)

If anything I've corrected yours.195.64.95.116 15:54, 31 Dec 2004 (UTC)

It may help here to consider a fact as something that is not known to be disputed anywhere today by otherwise reasonable people. Since there is obviously some dispute here about this "fact", our WP:NPOV policy arises to meet the occasion. Perhaps it is time to "characterize the dispute" if necessary, or merely to to back off on the fact with something like, "many people even claim they cannot detect a difference between P and Q." I will watch this page for a while. Be sure you are well familiar with the contents of the WP:NPOV article. And as always, remember wikilove and have a nice day! Tom - Talk 16:35, 6 Oct 2004 (UTC)

any criticisms toward the technology would be personal points of view made by the users of mp3's.

This makes no sense. Quality of audio-reproduction can be measured. If this wasn't the case, something like MP3 wouldn't exist, nor would it sound as good as it does these days.195.64.95.116 15:54, 31 Dec 2004 (UTC)

as they are based on the opinions of persons they should not be included in the article. instead try adding links to reveiws made by some sort of professional organization and let reader form their own opinions instead of trying to guide their opinions with your own through the artical--Larsie 21:40, 19 Oct 2004 (UTC)

Compression scheme vs. encoding scheme

CDDA does not compress audio. It is uncompressed 44.1 kHz, 16-bit stereo audio. CDDA is merely a format for encoding this audio along with error correction. MP3, on the other hand, is a compression format which can compress many sample rates and sizes, including 44.1/16/stereo, as well as a wide range of other combinations. Such as 22 kHz, 48 kHz, and even 96 kHz. Because of this, CDDA and MP3 are not directly comparable. You simply CANNOT say that one is better than the other, it's apples and oranges. Rhobite 21:42, Oct 27, 2004 (UTC)

Agreed, you cannot simply say that one is "better" than the other. But this is not because CDDA does "encoding" while MP3 does something completely different called "compression", but because the domain of MP3 is much greater than that of CDDA. Both are formats for storing PCM digital audio, and from the experience of the average computer user, both are overwhelmingly used for nothing except 44.1/16/stereo. For such an appropriately restricted application, they are perfectly comparable. One might say "using equipment XYZ and audio samples PQR, 50% of sample of 100 untrained listeners considered MP3 (using encoder MNO with settings JKL and bitrate ABC) to be not noticably worse than its CDDA source", and this would be a perfectly valid, reproducible test. [[User:Smyth|– Smyth]] 11:53, 30 Oct 2004 (UTC)

Fact remains that one can encode to MP3 from a much higher (high-end) quality audio source than the "box" where CDDA needs to fit in. For CDDA one would need to bring quality down, where samplerate, dynamics and bit-depth are concerned. This is not related to "tests" or opinions, this is sheer reality. CDDA has limitations as well, and they can actually be regarded as more important (reproduction-quality wise) than those of MP3. All this as long as it concerns playback quality and nothing more. 195.64.95.116 16:24, 31 Dec 2004 (UTC)

Sorry to resurrect this ancient post, but CDDA only supports one samplerate and bit-depth. If it isn't 44.1khz 16 bit, it isn't CDDA Nil Einne 15:31, 11 January 2007 (UTC)

Sampling rate

The text refers to "available sample frequencies." Would it be possible to define what "sample frequences" or "sample rate" means?

Sampling frequency? [[User:Smyth|– Smyth]] 01:46, 8 Nov 2004 (UTC)

Value judgments

Wikipedia articles don't make value judgments or recommendations, such as recommending that non-professionals never have a reason to use lossless compression. In any case, this statement is not true: "Those who will only listen, do not need to use lossless compression, since they won't hear the difference with MP3." You can't make blanket statements like that, some people can indeed tell MP3 - even with a good compressor - from audio that has never been compressed.

They can up to a measured degree. This can be proven and it often has been. Beyond certain high enough bitrates NO HUMAN will be capable of telling the difference.195.64.95.116 16:16, 31 Dec 2004 (UTC)

Etree and archive.org distribute lossless copies of nearly every show - there is obviously a large group of listeners who feels MP3 is inadequate for their uses. Rhobite 23:06, Nov 15, 2004 (UTC)

The fact that MP3 is now often considered a degraded format, surpassed by other formats, has nothing to do with that. It's simply because

1) there are bad mp3 encoders around (lots of them in fact)

2) MP3 tends to need some type of special treatment beforehand to reach optimum quality, other f

Yes, it's true that people under stress, with bad music equipment or in a noisy or unsuitable environment can have trouble differentiating between mp3 and lossless, or don't care, but that doesn't mean that they still can't hear it. Lossy formats should not be recommended with a clear conscience. BKmetic 00:00, 23 July 2006 (UTC)

This may be completely off kilter but to me it would seem simpler to actually look at the 5 difference in the signal given the identical encoder and differing bit rates. Any time someone uses "excellent" "acceptable", "poor" and "good" are all opinions (as oposed to factual). I personally have tried to hear the difference between the 128 bit and the CD on a good systems and I can hear the difference (after carefully selecting the music and carefully listening). In my car I have a great stereo, that would not be the case there since the road noise etc interferes just enough that I am unable to discern the difference. I will not make a value judgement as to what is acceptable, good or excellent since my opinons would only hold true to me at this instance. If one could measue it objectively it would be much more helpful. Look at it as follows Break down the audio into 5 frequency bands. Weight each band as it impacts our hearing (the center 3 bands are more important for human hearing than those below say 400hz and above 6kHz <--- Example) Using a digital oscilloscope, measure the difference in peaks in each band, then try to determine the loss of the resonant frequencies as opposed to the main frequencies since they tend to give music that "depth" the audiophiles like (yes, that is a qualitative opinion, not measurable but I am trying). Doing so would create a simple, objective comparison of formats and you could also use the data to compare encoders and thus determine the quality comparatively (objectively) in lieu of throwing subjective valuations around. [Moto] --71.112.37.172 19:28, 1 August 2006 (UTC)

Your simple view of what it takes to make objective quality measurements on audio signals is uninformed by the years of engineering effort in this direction. It is a hard problem, partly because hearing is so complicated, and quality measurements need to correlate to what you hear. Dicklyon 19:53, 1 August 2006 (UTC)

Minor Tidbits to Cleanup

The link in the "see also" section to Marcy Playground's music album "MP3" belongs in disambiguation. 68.100.224.150 17:56, 29 June 2006 (UTC)

"It provides a representation of PCM"

This isn't true. MP3 represents frequency domains, it can be decoded to PCM or DSD for example or in theory directly to analogue.

PCM

(From User talk:Smyth:)

You recently reverted my edit on MP3. MP3 does not contain a representation of PCM. It contains a representation of audio in a completely different way to how PCM represents audio. You can decode MP3 to DSD. It is NOT compressed PCM like FLAC is. Since you reverted my edit, I suggest you do some more research and then revert it again once you have learnt. Thanks. -- 82.152.177.71 13:11, 16 July 2006 (UTC)

I find it hard to get a definite answer for this because PCM is so universal, but references to each MP3 frame encoding exactly 1152 samples, and even being marked with a PCM sampling rate, make it sound very much as if PCM has a special status. Compare JPEG files: they contain a representation of an image in a completely different way to a straightforward rectangular grid, but they nevertheless are a representation of a rectangular grid of pixels, and not something else. – Smyth\^talk 15:22, 16 July 2006 (UTC)

If you don't know *what* PCM is, why on earth are you reverting edits related to the technical details of PCM? There are three essential ways to encode audio data (only two of which are widely used), and PCM simply uses an entirely different method than MP3.

Your analogy is not a very good one in this case. Horses and cars are the same because they can both be a means of transporting a person from one location to another, but that doesn't change the fact that they are very different on several fundamental metrics.

The specifications for PCM audio are available in several places on the internet, and they are generally technical (as is the case with most worthwhile specifications). Google is your friend, and if you don't understand the material then refrain from changing the information in the article.

Remove NPOV and replace with Unreferenced template?

i propose that the NPOV tag get removed and replaced with a {{Unreferenced}}. I would do it myself but ill be there might be disagreement which i think should be discussed. The only bit i think to be potentially not NPOV is the "Layer 1: excellent at 384 kbit/s, Layer 2: excellent at.. etcetc." part. Anyhoo. -(chubbstar) ^{— talk | contrib | 15:57, 26 July 2006 (UTC)}

Comments on encoders

I absolutely agree with this part :

>>Good encoders produce acceptable quality at 128 to 160 Kbit/s and near-
>>transparency at 160 to 192 kbit/s, while low quality encoders may never reach
>>transparency, not even at 320 kbit/s. It is therefore misleading to speak of
>>128 kbit/s or 192 kbit/s quality, except in the context of a particular
>>encoder or of the best available encoders. A 128 kbit/s MP3 produced by a good
>>encoder might sound better than a 192 kbit/s MP3 file produced by a bad
>>encoder. Moreover, even with the same encoder and resulting file size, a
>>constant bitrate MP3 may sound much worse than variable bitrate MP3

—Preceding unsigned comment added by Eclipsed aurora (talk • contribs) 30 July 2006

Would like to add the following comment to MP3 format performance for low bit rate. I use those bit rates to encode speech, and this comment comes from personal experience:

>> Low MP3 bit rates (64 bits mono and under) while adequate for speech

>>encoding, actually offer less compression than a straight mono .WAV

>>file at the same rate. The difference can be as much as 2 to 1.

>>

>> On the other hand, .WAV files at such low rates usually are not

>>supported in hardware-only players (non-computers). Also, the ability

>>of using ID tags is lost as well.

>> 24.181.71.243 16:23, 31 October 2006 (UTC) Alex

Licensing

Pardon me, but MP3 is already public domain. It is a file format. You cannot patent a file format as far as i know. that would be like Adobe saying they have a patent on PDF or Microsoft saying they have a patent on WMA or DOC files. You simply cannot patent or copyright or in any other way protect your idea if it involves a format of storing data. Just as you cannot patent things like NTFS or FAT or HFS, you cannot patent MP3.

MP3 is much more than a file format; see MP3. Encoding or decoding may use technologies that are patented, like Forgent saying they have a patent on JPEG (now ruled invalid), or Microsoft saying they have a patent on FAT filesystems (which they do, if those filesystems include long filenames). Dicklyon 01:33, 4 August 2006 (UTC)

24.181.71.243 16:36, 31 October 2006 (UTC) Alex I understand both your arguments. The point of the first comment, however, is that, in this context, it is meaningful to speak of the MP3 file format as a separate (though related) entity from the MP3 encoder itself, which is under patent. If I am not mistaken, the encoder design is patentable, but the output stream is not. It is conceivable that some other encoder design, known or unknown, may produce an MP3 compliant stream without resorting to the Fraunhoffer algorithm(s).

As the second comment points out, the difference may or may not be meaningful depending on context. 24.181.71.243 16:36, 31 October 2006 (UTC) Alex

Royalties are payable for commercial distribution of MP3 encoded content (http://www.mp3licensing.com/royalty/emd.html). No mention of this is made in the licensing section as it addresses only the implementation of encoders/decoders. Perhaps there should be mention of this? I believe there is no such royalty for the distribution of AAC content. - David (195.219.38.178 13:55, 27 July 2007 (UTC))

MPEG I, II, II.5

In my CDex program I have the option to encode with MPEG I, II oder II.5. Which one should I select? 84.60.102.161 08:38, 28 August 2006 (UTC)

Choose MPEG I. II and II.5 are for low bitrates e.g. speech.

BBC links Wikipedia's MP3 page

http://news.bbc.co.uk/2/hi/technology/5312696.stm --Daraheni 03:41, 5 September 2006 (UTC)

MP3 in Wikimedia projects

Why is the MP3 file format not used in Wikimedia projects? --84.61.39.144 14:47, 9 October 2006 (UTC)

patent issues. Vorbis is patent-free.

Text error- 18/11/06

Gerhard Stoll (Germany)) changed to Gerhard Stoll (Germany) (in the history section, paragraph 3 line 1)--Junkbot44 14:47, 18 November 2006 (UTC)

No Advantage?

I removed:

• the fact that these alternatives do not generally provide a clear advantage over MP3

Ogg Vorbis claims "For a given file size, Vorbis sounds better than MP3." [1]. If you put that sentence back, temper it (e.g. "the advantages are disputed") and provide citations in favor of and in opposition to Vorbis. Superm401 - Talk 09:33, 9 December 2006 (UTC)

I don't think the advantages are seriously disputed, but would not be significant enough to most users until they gave a file size reduction of more than a few tens of percent. – Smyth\^talk 10:49, 9 December 2006 (UTC)

Licensing and patent issues

A page I've been working on, list of software patents, refers to one of the MP3 patents, but did not cite sources. I came here trying to find sources, and have found lots more unsourced statements. Since the issue of software patents is so controversial, I think the things I've highlighted in the article need proper sourcing.

A prime example is the statement that development of MP3 was hindered by the patents, and yet the subsequent statement that it was still very popular. The success of the iPod doesn't, to me, indicate that development of MP3 has been hugely hindered!

I also deleted the final paragraph in the article because it made no sense. Patent applications filed in 1997 cannot "cover MP3" since MP3 was known by that time. Such patents might be able to cover extensions to MP3, or new variations on the format, but people cannot start getting new (valid) patents on it. Thomson have a limited number of patents listed on their website and it would be a relatively straightforward matter to work out when the last of those would lapse.

I don't know much about this topic, so can't really help in tidying up, I'm afraid. GDallimore 18:23, 3 January 2007 (UTC)

"Free Software" codecs

Ogg Vorbis and Speex were previously listed as being "free software codecs". This isn't really correct since they're actually just patent free, making them friendlier to free software, but not limited to being free software. People are still allowed to write proprietary software that uses these codecs. - James Foster 03:12, 5 January 2007 (UTC)

Actually they just believe they're patent free, no one knows for sure. The codec specification obviously isn't free software (how can a specification be software?) but the only implementation of the specification we know of, i.e. the codec, is in fact free software. If someone wants to write proprietry software they will have to fulfill the terms of the codec license OR write their own implementation of the specification, i.e. a new codec supporting Vorbis or Speex Nil Einne 15:26, 11 January 2007 (UTC)

The article first says that the double-blind testing had five schemes tying for first place. Then later it says that "some claims" that Ogg Vorbis is superior to WMA and others. Given the strong political context of free software vs. a Microsoft standard, I think its important for the article to be very NPOV and avoid unsubstantiated claims about relative quality.

Role of AT&T (Bell Labs) in the development of MP3?

The role of AT&T/Lucent/Bell Labs in the development of MP3 is missing in the article. According to this source (dated 2007-02-16),

AT&T Corp. and Fraunhofer agreed in 1989 to develop MPEG-1 Audio Layer 3 technology, now called MP3. Scientists from AT&T's Bell Labs collaborated with Fraunhofer before AT&T spun off the unit in 1996. Bell Labs became Lucent Technologies Inc., which Alcatel SA acquired last year.

Another source (dated 2007-02-23) informs:

What was Alcatel-Lucent's role in developing MP3?

The MP3 technology was developed in large part by people with Germany's Fraunhofer and AT&T's Bell Labs, which became part of Lucent when it was spun off in 1996. Alcatel and Lucent merged last year, becoming Alcatel-Lucent.

-- HYC 05:38, 24 February 2007 (UTC)

I have edited parts of the article, added references to Bell Labs work, as well as mentioned Thomson-Brandt, who is to be credited with the window-switching understanding, in parts of the article.

I am unfamiliar, to date, with the actual process involved in editing Wikipedia, so please forgive me if I do not note some particular standard of notation, etc.

Woodinville (talk) 23:09, 15 January 2008 (UTC)

I did some more cleaning up in the history and "suzanne vega" area. Pointed to Fletcher's work that Zwicker built on, removed some personal point of view, and tried to sort out some grammar. Woodinville (talk) 21:12, 24 January 2008 (UTC)

I'm thinking about moving a bit of the history section from this article into the audio compression article [2]? Any suggestion/opposition? --Gabriel Bouvigne (talk) 22:06, 24 January 2008 (UTC)

Not particularly. It would be good if the present article remains balanced, but the section has perhaps grown a bit much. (looks at the article) My goodness, that article needs an exposition on perceptual vs. source coding, as well, doesn't it? Just, if you will, try to keep the coverage here evenhanded, unlike its history up to last week.--Woodinville (talk) 08:36, 25 January 2008 (UTC)

I just found that we have a page about perceptual audio coding [3]. So my idea would be:

• have explanations about perceptual vs source coding within the audio compression article [4]
• Transfer most of the psy-coding history from the mp3 article into the perceptual audio coding article [5], and extend it to cover the early PXFM and OCF, Musicam, PAC/EPAC, AAC and the likes, so we could have a central place to clearly show the timeline and contributions
• Link the various existing coding schemes articles (mp2, mp3, aac, ac3,...) to the perceptual audio coding article.

This way we could perhaps finally end up with something really informative. (of course, this will probably require some time)--Gabriel Bouvigne (talk) 15:33, 25 January 2008 (UTC)

not an ipod

Yes, I'm being controversial here, but since the top picture is of a non-ipod mp3 player, which nobody has and nobody can identify, why not change the pictures caption to "an (atypical) mp3 player!" --81.105.251.160 16:08, 1 April 2007 (UTC)

Development of MP3; role of OCF and ASPEC

The statement in paragraph 1 of the Development section: "Modern lossy bit compression technologies, including MPEG, MP3, etc, are based on the early work of Prof Oscar Bonello of the University of Buenos Aires, Argentina." is erroneous. The psychoacoustic masking codec was first proposed by Manfred R. Schroeder et al. in the Journal of the Acoustic Society of America, Vol. 66. pp. 1647-1652, "Optimizing Digital Speech Coding by Exploiting Masking Properties of the Human Ear", in 1979.

Paragraph 4 states "In 1991, there were two proposals available:", but neglects to mention the other proposal, ASPEC.

In general the development of MP3 was via the progression OCF(1988)->ASPEC(1991)->MP3(1994), with a small contribution from Musicam. This progression is distorted or not present at all in the article. The article overemphasizes the contribution of Musicam. MP3 is essentially a transform coder derived from ASPEC and OCF. Musicam was a subband coder. The only contribution of Musicam to MP3 was the division of OCF into two transform-coded subbands.

Without objection I will edit the article accordingly.

William spurlin 18:51, 18 June 2007 (UTC)

Bonello contributions factually inaccurate

I am unaware of any contribution in the electrical engineering, broadcasting or acoustical literature by Oscar Bonello. In particular the claim that "the world's first bit compression system" was developed by Oscar Bonello is absurdly at variance with the actual devlopment of such systems as vocoders and adaptive differential pulse code modulation. Unless the author of the Bonello section can factually document his/her claims, I propose its removal.

William spurlin 01:12, 23 July 2007 (UTC)

The only reference I can find to this claim on the internet is this link to a company that Bonello apparantly started back in the day. http://www.solidynepro.com/indexahtmlp_Hist-ENG,t.htm

85.24.231.191 11:29, 5 September 2007 (UTC)

This is painfully inaccurate. I'm sure a good detective could find bit compression dating back to the 1950s, but United States Patent 4117470, filed Oct 8th, 1976 predated Bonello by over a decade. GPS systems could be said to use bit compression and they were designed in the late 1960s. I will remove this entry. —Preceding unsigned comment added by 67.113.109.220 (talk) 05:13, 30 October 2007 (UTC)

You could also cite Johnston, J. D. and Goodman, D. J., “Multipurpose hardware for digital coding of audio signal,” Proc. NTC77, 1977. which cites work done in 1975 summer and 1976 summer, culimating in flexible hardware doing ADPCM (or APCM) at rates from 2 bits at 8kHz to 12 bits at 37 kHz (the odd number due to the speed of the A to D). The higher rates were certainly music compression, indeed, although of an extremely primitive kind.

Woodinville (talk) 23:52, 15 January 2008 (UTC)

Ok, I see this claim about Bonello is back. Johnston gave a talk in 1988 at the Mohonk Conference, Krasner and Krahe are both earlier than that, and Manfred Schroeder et al comes from publication in 1979. Given that, claiming priority with a 1989 publication (arriving after the JSAC articles nearly everyone on the planet who did this stuff) is simply nonsense. Woodinville (talk) 21:15, 1 February 2008 (UTC)

Asked OscarJuan for clarifications --Gabriel Bouvigne (talk) 13:30, 2 February 2008 (UTC)

Enough is enough here. Bonello is cited as 1989 here, but as 1987 on the audio data compression page. Both 1989 and 1987 are completely uncited, and I can't find anyone off-line who knows a single thing about this work. It's time for this to either appear in fully supported, testable, verifiable form, or for it to be regarded as vandalism. What's more, the priority claim back to 1983 is no more supported than the priority claims of anyone else before publication, and the maturity of the algorithms in the JSAC paper makes it clear that lots of people were working on this in 1983. The claims here are inconsistant and completely unsupported. Barring full support, they should go away. Compare this to the AT&T claims, for instance, that are absolutely supported by both patent and publication, ditto for OCF, etc. Woodinville (talk) 22:58, 5 February 2008 (UTC)

I have reverted the latest attempt at "Bonello Contributions". Nobody but Oscar Juan seems to knwo about them in a fashion that can be tested and verified. He claims "the first" when an entire book full of fully developed algorithms is published a year before his own claim. This is unuspportable. Woodinville (talk) 23:08, 15 March 2008 (UTC)

Why Bonello is considered the inventor of bit compression technology ?

In first place I will analyze the unfortunately comment of Mr Spurling: ''I am unaware of any contribution in the electrical engineering, broadcasting or acoustical literature by Oscar Bonello'' Yes, I know that some persons do not known about Aristotle and his work, maybe never enjoy the Beethoven 9 Symphony or never read to Proust. Probably he do not read in Spanish or German. Probably he do not read ASA or AES Journals, etc. But the very estrange situation is that placing my name at Google he will found valuable information and hundreds of citations. With 25 years of teaching at the University of Buenos Aires (4 Nobel Prize, 180 years activity) I have 150 published papers and are Fellow of the Audio Engineering Society, New York Please see my CV at: http://www.solidynepro.com/documentos/Bonello-English%20CV.doc

Some of the above mentioned authors have a confusion between the concept of theoretical work and the concept of "realization"; give birth to a new technology. By example mathematician Euler in 1750 works creating algorithms for a rocket that fly to space. Fourier and Laplace works with mathematical approaches to this type of problems. A lot of people helps to create the scientific bases of a machine that goes outside the earth gravity... All wonderful, but none of them were able to flight... But a reduced group of American people in 1969 reaches the moon surface... Of course it was the work of a team (Newton and Euler included), but the space era starts in 1969 and not when Euler found his algorithms.

The confusion is between delivering a paper or fill a patent, and on the other hand, create a new technology that works and serves the human been.

Studies about masking of bands started in 1924. Although Dr Helmholtz at his 1885 publication demonstrates to know this ear property (only explained after 1960) Then, the previous work of the foundation of a science is very important... but this do not means that the real thing will start working. (Remember all the airplanes designed by the genius of Leonardo da Vinci; but... Leonardo was not the inventor of the aviation, the Wright Brothers did it.

Bit compression technology is very similar. A lot of people works with this idea (If we include Dr Helmholtz, since 1885). But one thing is to create the theoretical bases and a different thing is to create the final working device; the invention.

Our developing team in Argentina had 3 challenges 1) Develop a bit compression algorithm // 2) Create an audio car that cab perform in real time de coder and decoder of the audio streaming // 3) Develop the automation software that ran at the early IBM PC XT machine in order to produce high quality audio signals Since our university de no have founds to support this project, we use a private support with two conditions: a) To get a bit compression system of FM audio quality running on a PC, with reliable operation and ready for commercialization. b) All the research and developing must be done in secret, no papers, no patents, because the sponsor wishes to be the first one to start this technology

We was carefully to have a perfect documentation to avoid any doubts in the future. First presentation in Argentina at the Secretary of Communication in front of 150 engineers. Presentation at the NAB Radio Show in 1990 (All American radio technician were there ! (Do you need more proof ?) Then, news at newspapers, magazines, Advertising at the AES Journal, Canadian, Spain, France, magazines. Installation in radio stations (KIKO AM in San Francisco, California was the world first in 1990. Installations in Radio France, Radio Finland at Oulu, and lot of radio stations around the world.

All is perfectly documented. Of course I am glad to give more information at oscar@solidyne.com.ar and newspapers copy, advertising, client list, etc and contact IEEE engineers that known in deep this technology. Please note we are NO presenting an idea or algorithm We are NOT presenting audibility curves We are not presenting block diagrams or circuits... We present a real working device, Like the Wright Brothers, Like Graham Bell, Like the Edison lamp... —Preceding unsigned comment added by OscarJuan (talk • contribs) 04:03, 16 March 2008 (UTC)

Oscar Bonello considered the inventor of "bit compression technology"? I'm sorry, but isn't this a bit too broad? Anyhow, if you are the recognized inventor of bit compression, then for sure someone else will mention it within the relevant articles, and so there is no need to add yourself this section about you, isn't it? --Gabriel Bouvigne (talk) 22:53, 16 March 2008 (UTC)

Gabiel: I agree with you (sometimes it happens...) that it sounds "oversized" I made several corrections to focus on the real practical invention (the concept of "The first working device" using bit compression) Please read it. I am open to change the text in order to get a description that shows exactly what we have done and at the same time have full agreement with the MP3 society opinions

PS: Is unfair your comment about "someone else will mention it " You know my CV and know several national Prizes we have about it... Of course "invention" is a difficult task (today 150 years later there are a lot of telephone inventors in France, Germany, England, Italy, Russia, etc) (Oscar Bonello, March 16) —Preceding unsigned comment added by OscarJuan (talk • contribs) 00:33, 17 March 2008 (UTC)

I think that it's time that Oscar Juan stops stealing credit from others. Krasner had a working realtime device that was published. Brandenburg had OCF in real-time form, just to mention two examples.

You are welcome to assert that you were one of the contributors. You are not, however, in any fashion, "the first", and your claiming so is unacceptable. Please adjust your claims immediately. You are demonstrably, provably not the first. OCF alone refutes your claim by any evidence available to me.

Furthermore, it is purely disingenious to appear authoritive by citing the standard psychoacoustic works that we all use in our daily work. What's more, it would be good of you to cite the basic work on masking that was demonstrated by Fletcher, et al, although certainly not in the modern understanding, the applicability of equal loudness curves, how they interact with "Weber's Law" and the like. You may well have contributed, but you are NOT the only and sole inventor, sir, and you are stealing credit from a wide variety of other individuals when you claim otherwise. Correct your claims. Woodinville (talk) 01:17, 17 March 2008 (UTC)

Considering that Bonello's Audicom/ECMA system is based on psychoacoustic masking, in the same way as MPEG audio, but without MPEG audio being based in any way on ECMA, and also considering that ECMA was not the first psychoacoustic lossy encoding scheme, that means that ECMA may be part of the same familly as MPEG audio, but is clearly not an ancestor. Thus, it should not be claimed to be such an ancestor, but rather should be acknowledge to be the first radio automation device using those techniques.

That means that it should be placed within the "Audio compression (data)" and/or "Broadcast automation" articles, and is irrelevant to the mp3 article. Please note that it is already properly described within those articles, and thus should be removed from the mp3 article. --Gabriel Bouvigne (talk) 11:28, 17 March 2008 (UTC)

Bonello's reply to Mr Woodinville and Mr Bouvigne

a) The works of Kramer, Brandemburg and others are not "practical devices" Only valuable Lab experiences, without further applications in real life and the everyday work. Then we can not consider it "inventions" The same way Leonardo did nice drawings of flying machines, but he is not considered the "inventor" of the airplane. b) We agree to limit claims and to move the long part to Broadcast Automation. Then, we did a short two lines text with very limited claims. I hope you agree with it. c) Gabriel do not like references to the acousticians that develops the principles of the ear masking. Then, who reads the MP3 article believes that all starts with a small group of people involved at the MPG Project. I think that you do not change the History and recognize the early contributors. If you will, please place the names at the beginning of the MP3 article. If not, is not bad that Bonello remembers them. c) As I told you earlier, Gabriel, since "MP3" is not only a technical expression because usually the people associates the name with "sound from a PC" I understand that is correct to have a small mention (now very short) to the first PC working system used now up to this days, Best regards- OscarJuan —Preceding unsigned comment added by OscarJuan (talk • contribs) 20:27, 20 March 2008 (UTC)

COMMENTS ABOUT Woodinville, Gabriel and OscarJuan discussion

Bonello is rigth when he said that he is the inventor of the first working bit-compression device. I assist at the first Audicom presentation in NAB 1990, Atlanta, USA and at that moment nobody at the world is able to offer somtehing similar About the Woodinville unfortunately comment to profesor Bonello of "disingenious" I feel this is unfair and agressive (please remember,sir, we are not dancers at a cabaret...) I think that is correct to recognize the people who advance the science knolewdge. From what we know, the Bonellos's work was based in Richard Ehmer masking curves, not at the Schroeder analytical approximation used in MP3 (that do not fit exact with the real ear masking curves, as you can easily see comparing both) In my personal opinion the MP3 page gives false information when starts the technology at 1979. Then I undestand that you do not agree that Bonello gives the name of the true precursors. I encourage you correcting the false impression that "all was done by a group of a few good guys starting in 1979..."

Roberto miller (talk) 16:40, 22 March 2008 (UTC) Roberto Miller Roberto miller (talk) 16:40, 22 March 2008 (UTC)

First, it is incorrect to argue that Schroeder analytical approximation has anythign to do with masking curves, it is an approximation of the shape of a cochlear filter bank. On this statement alone, we can discount the comments from Miller. Certainly the authors were well aware of this approximation, and well aware at the time that it was not a "masking curve". Second, the dismissal of OCF, etc, by previous comments above is likewise improper, and the nonsense about "only valuable lab experiences" is simply atrocious. Many people other than Bonello had lots of valuable lab experience and lab experience that is the basis from which all of the above work, Bonellos and others, comes about. The entire subject is based on LABRATORY EXPERIENCE. The AT&T contributions, for instance, date back from Fletcher right through the authors of the MP3 standard. Calling those "valuable lab experience not real life" is quite inaccurate. It is completely impossible to know what some others have in the way of "valuable lab experience not real life", therefore making such claims is both offensive and unjustified. Science is testable, verifiable, and repeatable. Lab experience is a way to test, verify and repeat, therefore dismissing it is inappropriate. It is not acceptable to continue to claim priority with untestable claims, and it is likewise inappropriate to dismiss the acknowleged work in the field as "not real life". Finally, I've pointed to Fletcher, a pioneer in auditory research that much of Zwicker's early work was based on, and others have pointed out a variety of other psychoacoustic results, so the claim that they are unacknowleged, along with the profuse name dropping from Oscar Juan, is just absurd. Woodinville (talk) 05:53, 25 March 2008 (UTC)

ABOUT LAB EXPERIENCES

Mr Woodinville insist in not understand the difference between a "Lab experience" and a technical invention. Very far of my idea is dismissing a lab experience, as Mr Woodinville believes. I know the importance of Lab experiences because I teached 25 years Theory of Sound and Psychoacoustics at the University of Buenos Aires. My name is associated with the Bonello's Criteria for natural modes in room acoustics (please see BONELLO CRITERIA - Págs. 56 - 58 of the book Handbook for Sound Engineers - Glen Ballou (Howard Sams) - USA or THE BONELLO CRITERIA - Págs. 110-112 of book: Acoustic Techniques - Alton Everest (TAB Books) - USA) This worldwide used acoustic criteria born in "lab experiences" Then please accept that we agree with you that "Science is testable, verifiable, and repeatable. Lab experience is a way to test, verify and repeat..."(sic) But you must understand that the "invention" is a different thing.. By example: a) Leonardo did nice drawings of flying machines, but he is not considered the "inventor" of the airplane; b) Heron (Greek about 250 BC) created a "lab experience" to demonstrate how the steam can move a machine. But the Industrial Revolution needed the Watt invention to have true machines. c) The excellent Helmholtz analysis of the voiced sounds and the excellent lab experiences he creates (On the sensations of Tone) is the basis for telephone communication. But the telephone inventor was Graham Bell... Please note that Leonardo, Heron or Helmholtz are not associated with the "invention" of Airplane, Steam Machine or Telephone. Please note that this is a fact and not a "improper, and the nonsense " as you stated. I hope this comments will help you to understand the difference between a lab experience and a real life invention. I do not (as you suppose) say to one thing is better than the other. My statement is only: there are different things ! About your comment "It is not acceptable to continue to claim priority with untestable claims" I can give you a lot of testable information if you wish (I did it with Gabriel) Please give your mail address. The mine is oscar@solidynepro.com Regards OscarJuan (talk) 01:00, 27 March 2008 (UTC)OscarJuanOscarJuan (talk) 01:00, 27 March 2008 (UTC)

Mr. Bonello, let us meet at NAB and discuss this in person. You are very wrong, and you are completely misrepresenting others' work.

71.231.2.119 (talk) 19:54, 27 March 2008 (UTC)

Ok, I've looked around the show floor. I can't find Solidyne anywhere, and I still haven't any testable evidence of Mr. Bonello's priority for invention of audio coding, in fact, all of the available literature puts any number of people (Schroder et al, Krahe, Krasner) before him. The obfuscation about "what is invention" here is an argument that simply avoids the actual literature, patents, and other evidence. —Preceding unsigned comment added by Woodinville (talk • contribs) 23:23, 15 April 2008 (UTC)

Ok, let's get this straight. Bonello is not, repeat NOT the inventor of perceptual audio coding. All of the arguments above are simply a kind of offensive hand-waving that steals credit from others.

HOWEVER, you will all please note that I DO give Bonello credit for early broadcast automation. I see no objection to that, and I do see evidence that he was there very, very early in that cycle. Woodinville (talk) 19:46, 28 April 2008 (UTC)

Misinformation on the chapter explaining the structure of an mp3 file

The picture on the article claims, that the header starts with value [1111,1111,1111] followed by bit marking the MPEG version [1]. Apparently the correct form is: [1111,1111,111] followed by two bit sequence marking the MPEG version ([00] = MPEG version 2,5; [01] = reserved; [10] = MPEG version 2; [11] = MPEG version 1).

I noticed this while working with a project where I need to split an MP3 file. And I had trouble finding correct headers :-)

I also found a link explaining the anatomy of the header more deeply.

http://www.mp3-tech.org/ —Preceding unsigned comment added by 212.50.128.114 (talk) 06:55, 4 October 2007 (UTC)

ack. moved to proper place.

Woodinville (talk) 23:49, 15 January 2008 (UTC)

Expiration of Patents.

Removed:

However, U.S. patents can only last up to 20 years from the date of filing which must be within one year of first publication. MP3 was released as a specification in 1991, so if U.S. courts applied U.S. law, no patent claims could apply to MP3 itself after 2012.

Any U.S. patent filed after 1992 should (by law) have any claims pertaining to MP3 struck down considering the published specification as prior art. If they had been published earlier (such as in public drafts), the latest date would be even earlier.

Current US patent law has the patents expire at most 20 years after filing. However, at the time that many of the patents were filed, it was 17 years after the patent was granted. In the big list of patents, #5,924,060 was filed August 29, 1987, so it was filed before the published specifications. But it was granted on July 13, 1999, so 17 years after that is July 13, 2016. So, sorry, but under the law that it was granted under, it expires in 2016. Jrincayc 03:23, 5 October 2007 (UTC)

See [6]. The law changed in 1995, so presumably patent 5,924,060 was under the old law of 17 years from being granted. Jrincayc 03:27, 5 October 2007 (UTC)

Has anyone done a study of when the patents required for playing an mp3 expire? Jrincayc (talk) 04:48, 4 March 2008 (UTC)

I could be wrong, but it looks like it might be 2013 is the year that the last patents for playing an MP3 expire. I am assuming that anything filed after 1993 must only be necessary for encoding, since MP3 was a standard by then. All of the other ones that expire after 2013 look like they are about encoding MP3s. Can someone verify this? 75.174.1.63 (talk) 19:29, 27 April 2008 (UTC)

After looking it up myself, here are the facts that I found: The initial near complete MPEG-1 standard (parts 1,2,3) was publicly available in December 6, 1991 as ISO CD 11172. [7][8] The final version of the MPEG-1 specification describes decoding and includes pseudo-code for the decoding. It also has hints as to how to encode MPEG-1 audio, but is not as detailed for that. In the US, patents must be filed within one year of publication or they are invalid. US patents filed before about 1995 last the longer of 17 years from the date that they are granted, or 20 years from the date of filing. The last US MP3 patent to expire that was filed by December 1992 (one year after publication of the draft standard) expires in December 2012. So, it should be theoretically possible to implement MP3 decoders patent free at that time. In otherwords, I think that a statement similar to the one I removed can be readded. If no-one has any objections I will do so after at least a week, or someone else can do so. Jrincayc (talk)

I added it back into the article, with three reference links. Jrincayc (talk) 21:52, 12 October 2008 (UTC)

Either I copied the date down wrong, or it was wrong at the MP3 patent list, but #5,924,060 was filed on August 29, 1997, so it probably can't apply to MP3 decoding, since that was specified in the August 1993 ISO MPEG-1 standard ISO/IEC 11172-3. Jrincayc (talk) 13:42, 1 June 2008 (UTC)

Here is the list of when MP3 patents expire. I am not sure how to deal with a reissued patent, but I think it probably expires based on the original patent date. US 5,742,735 is the last one to expire that was filed by August 1994 (one year from the MPEG-1 decoding spec being published, so decoding probably is safe by April, 22 2015. Some of these patents may not be necessary for encoding or decoding MP3. Also, I just did the later of 17 after the granting date, or 20 years after the filing date. Jrincayc (talk) 14:49, 1 June 2008 (UTC) (Replaced with better table using User:Jrincayc/Patent_utils Jrincayc (talk) 03:37, 17 July 2008 (UTC))

Patent	Filed	Granted	Expiration	Summary	Notes	Company
5,341,457	20 aug 1993	23 aug 1994	20 aug 2013	Perceptual coding of audio signals	[9]	Alcatel-Lucent
RE39,080	22 sep 1994	06 may 1997	22 sep 2014	Rate loop processor for perceptual encoder/decoder	Reissue of 05627938 filed 13 aug 2002 granted 25 apr 2006 [10]	Alcatel-Lucent
4,972,484	21 jul 1988	20 nov 1990	21 jul 2008	Method of transmitting or storing masked sub-band coded audio signals	[11]	Audio MPEG, Inc
5,214,678	31 may 1990	25 may 1993	31 may 2010	Digital transmission system using subband coding of a digital signal	[12]	Audio MPEG, Inc
5,323,396	21 dec 1992	21 jun 1994	21 dec 2012	Digital transmission system, transmitter and receiver for use in the transmission system	[13]	Audio MPEG, Inc
5,539,829	07 jun 1995	23 jul 1996	07 jun 2015	Subband coded digital transmission system using some composite signals	[14]	Audio MPEG, Inc
5,606,618	27 dec 1993	25 feb 1997	27 dec 2013	Subband coded digital transmission system using some composite signals	[15]	Audio MPEG, Inc
5,530,655	06 jun 1995	25 jun 1996	06 jun 2015	Digital sub-band transmission system with transmission of an additional signal	[16]	Audio MPEG, Inc
5,777,992	07 jun 1995	07 jul 1998	07 jun 2015	Decoder for decoding and encoded digital signal and a receiver comprising the decoder	[17]	Audio MPEG, Inc
6,289,308	08 mar 2000	11 sep 2001	08 mar 2020	Encoded wideband digital transmission signal and record carrier recorded with such a signal	[18]	Audio MPEG, Inc
5,481,643	24 apr 1995	02 jan 1996	24 apr 2015	Transmitter, receiver and record carrier for transmitting/receiving at least a first and a second signal component	[19]	Audio MPEG, Inc
5,544,247	25 oct 1994	06 aug 1996	25 oct 2014	Transmission and reception of a first and a second main signal component	[20]	Audio MPEG, Inc
5,610,985	21 jan 1994	11 mar 1997	21 jan 2014	Digital 3-channel transmission of left and right stereo signals and a center signal	[21]	Audio MPEG, Inc
5,740,317	30 aug 1995	14 apr 1998	30 aug 2015	Process for finding the overall monitoring threshold during a bit-rate-reducing source coding	[22]	Audio MPEG, Inc
5,878,080	07 feb 1997	02 mar 1999	07 feb 2017	N-channel transmission, compatible with 2-channel transmission and 1-channel transmission	[23]	Audio MPEG, Inc
5,960,037	09 apr 1997	28 sep 1999	09 apr 2017	Encoding of a plurality of information signals	[24]	Audio MPEG, Inc
5,991,715	31 aug 1995	23 nov 1999	31 aug 2015	Perceptual audio signal subband coding using value classes for successive scale factor differences	[25]	Audio MPEG, Inc
6,023,490	09 apr 1997	08 feb 2000	09 apr 2017	Encoding apparatus for encoding a plurality of information signals	[26]	Audio MPEG, Inc
4,821,260	16 dec 1987	11 apr 1989	16 dec 2007	Transmission system	[27]	Thomson
4,942,607	03 feb 1988	17 jul 1990	03 feb 2008	Method of transmitting an audio signal	[28]	Thomson
5,214,742	01 oct 1990	25 may 1993	01 oct 2010	Method for transmitting a signal	[29]	Thomson
5,227,990	17 jan 1992	13 jul 1993	17 jan 2012	Process for transmitting and receiving a signal	[30]	Thomson
5,384,811	24 aug 1992	24 jan 1995	24 aug 2012	Method for the transmission of a signal	[31]	Thomson
5,736,943	31 may 1996	07 apr 1998	31 may 2016	Method for determining the type of coding to be selected for coding at least two signals	[32]	Thomson
5,455,833	26 apr 1993	03 oct 1995	26 apr 2013	Process for the detecting of errors in the transmission of frequency-coded digital signals	[33]	Thomson
5,559,834	15 apr 1994	24 sep 1996	15 apr 2014	Method of reducing crosstalk in processing of acoustic or optical signals	[34]	Thomson
5,321,729	26 apr 1993	14 jun 1994	26 apr 2013	Method for transmitting a signal	[35]	Thomson
5,706,309	02 may 1995	06 jan 1998	02 may 2015	Process for transmitting and/or storing digital signals of multiple channels	[36]	Thomson
5,701,346	12 sep 1996	23 dec 1997	12 sep 2016	Method of coding a plurality of audio signals	[37]	Thomson
5,742,735	25 aug 1994	21 apr 1998	25 aug 2014	Digital adaptive transformation coding method	[38]	Thomson
5,812,672	15 dec 1994	22 sep 1998	15 dec 2014	Method for reducing data in the transmission and/or storage of digital signals of several dependent channels	[39]	Thomson
5,579,430	26 jan 1995	26 nov 1996	26 jan 2015	Digital encoding process	[40]	Thomson
6,185,539	26 may 1998	06 feb 2001	26 may 2018	Process of low sampling rate digital encoding of audio signals	[41]	Thomson
6,009,399	16 apr 1997	28 dec 1999	16 apr 2017	Method and apparatus for encoding digital signals employing bit allocation using combinations of different threshold models to achieve desired bit rates	[42]	Thomson
5,924,060	20 mar 1997	13 jul 1999	20 mar 2017	Digital coding process for transmission or storage of acoustical signals by transforming of scanning values into spectral coefficients	[43]	Thomson
5,703,999	18 nov 1996	30 dec 1997	18 nov 2016	Process for reducing data in the transmission and/or storage of digital signals from several interdependent channels	[44]	Thomson

MP3

Sorry - I just need to say this: an MP3 is NOT the smae thing as a Mp3 player! It is so annoning when people say "can I see your mp3" and I'm like DRR, that's a file format not a music player! Even commercial sites do this! (if you want, add this into the main MP3 page :)

71.62.220.166 20:17, 11 October 2007 (UTC) Kyle

Hours of playback per GB

I recently was looking at the specifications for a 4GB MP3 player. It stated the following:

"4GB of memory plays back over 64 hours of MP3 (128 hours of WMA) music (over 960 MP3/1920 WMA songs)* (based on 128Kbps playback)"

Now this doesn't make any sense to me. If both the mp3 and the wma music is 128 kilobits per second, then shouldn't 64 hours of WMA take up exactly the same amount of space as 64 hours of MP3?! —Preceding unsigned comment added by 69.29.68.126 (talk) 03:41, 31 October 2007 (UTC)

Sorry to jump in so late. Yes. Absolutely. Somebody at that company's marketing department better have gotten a major spanking for that mistake. — NRen2k5^(TALK), 02:47, 11 March 2008 (UTC)

What does the "mp" in "mp3" stand for?

Anyone know? —Preceding unsigned comment added by 59.3.33.196 (talk) 04:36, 29 December 2007 (UTC)

Well MP3 means MPEG-1 Audio Layer 3 so i guess its MPEG. Sir Fritz (talk) 08:30, 6 January 2008 (UTC)

Since MPEG stands for Motion Picture Experts Group, the name of the team who developed the standard, and because MP3 is short for "MPEG-1 Audio Layer 3, it's safe to say that MP3 is more of an abbreviation than an acronym. It doesn't "stand for" anything. Does anyone think this should be addressed briefly in the article? Fungicord (talk) 08:50, 10 January 2008 (UTC)

The name obviously came from the extension .mp3, which itself is derivative of the .mpg used for MPEG-1 and -2 video files; the restricting factor is that MS-DOS and earlier versions of Windows were limited to a 8.3 file naming convention (IIRC this changed with Windows 95, I recall being amazed about finally being able to use descriptive file names such as "Letter to Mum about dirty laundry.1.doc"). Maikel (talk) 22:51, 6 March 2008 (UTC)

The scope and audience of the article

The word "MP3" today more often refers to an MP3 file than the MP3 encoding scheme itself. I think this should be made apparent in the beginning of the article. This article does a disservice to readers who don't already know what an MP3 file is. Just imagine a technologically-challenged grandmother who links to this article on Wikipedia from Google because her granddaughter just asked for an MP3 player as a gift. The article has so far started out way too technical. Sentence 2 said "lossy compression algorithm" and sentence 3 refers to bit-rates. This gets too technical too fast for such a widely known term. Maybe there should be another article named "MP3 file" that allows separation of the mainstream uses of the term as a file type from the technical details of the encoding scheme. For now, I'll add a couple sentences at the top of the article that are a little less shocking to non-technical readers. Fungicord (talk) 09:56, 10 January 2008 (UTC)

Actually, I think the article should be improved in the other direction: There is almost no in-depth discussion about the technology. There is some talk about the file format, but what the codec actually does to the audio it processes (e.g., the transform coding, the psychoacoustic model, the vector quantization) receives very little technical discussion, and these are in my opinion the most relevant parts of the technical discussion; what bitsequence appears at the beginning of each file is an arbitrary and not very interesting fact. After reading almost all of the article, I cannot say much more than a few buzzwords about how MP3 works. Compare, for example, Vorbis#Technical details. --Zvika (talk) 15:52, 5 February 2008 (UTC)

I also agree on the "more information" side. I must point out, though, there is no VQ in MP3. 66.78.202.165 (talk) 23:36, 11 April 2008 (UTC)

Name

MPEG-3 claims that MP3 means "MPEG-1 Part 3 Layer 3" and not "MPEG-1 Audio Layer 3". Which is right? Maikel (talk) 22:46, 6 March 2008 (UTC)

"MPEG-1 Audio Layer 3" is the one I always hear. It's possible that "MPEG-1 Part 3 Layer 3" is a more formal way of describing it, as to how it fits in the bigger scheme of things. Anyway, I edited the MPEG-3 article to mention MP3 as "MPEG-1 Audio Layer 3". — NRen2k5^(TALK), 02:55, 11 March 2008 (UTC)

Fraunhofer's contributions

FHG certainly deserves a lot of credit for popularizing MP3 as well as major technological contribtions, but this article is repeatedly revised to steal professional and technical credit from AT&T Bell Labs (Johnston), Thomson-Brandt (Spille, Schroder), and CNET (Mahieux).

FHG was not the sole inventor, nor does it deserve sole credit. Brandenburg, as already documented, was working at AT&T BELL LABS with/for Johnston at the time of creation of the standard, and was travelling on an AT&T Budget, along with Johnston, who also played a primary part in creation of the algorithm, as documented in the published psychoacoustic models. This is hardly a sole FHG activity. Please in the future, do not steal credit and slight people professionally. —Preceding unsigned comment added by Woodinville (talk • contribs) 23:29, 11 April 2008 (UTC)

Patent List

List of the MP3 Patents for reference. Jrincayc (talk) 15:03, 11 May 2008 (UTC)

• Alcatel-Lucent [45]
  • US 5,341,457 -- Perceptual coding of audio signals -- Filed: August 20, 1993 Granted: August 23, 1994 [46]
  • US RE39,080 -- Rate loop processor for perceptual encoder/decoder -- Filed: August 13, 2002 Granted: April 25, 2006 [47] Reissue of 05627938 Filed: Sep., 1994 Granted: May., 1997
• Audio MPEG, Inc [48]
  • US 4,972,484 -- Method of transmitting or storing masked sub-band coded audio signals -- Filed: July 21, 1988 Granted: November 20, 1990 [49]
  • US 5,214,678 -- Digital transmission system using subband coding of a digital signal -- Filed: May 31, 1990 Granted: May 25, 1993 [50]
  • US 5,323,396 -- Digital transmission system, transmitter and receiver for use in the transmission system -- Filed: December 21, 1992 Granted: June 21, 1994 [51]
  • US 5,539,829 -- Subband coded digital transmission system using some composite signals -- Filed: June 7, 1995 Granted: July 23, 1996 [52]
  • US 5,606,618 -- Subband coded digital transmission system using some composite signals -- Filed: December 27, 1993 Granted: February 25, 1997 [53]
  • US 5,530,655 -- Digital sub-band transmission system with transmission of an additional signal -- Filed: June 6, 1995 Granted: June 25, 1996 [54]
  • US 5,777,992 -- Decoder for decoding and encoded digital signal and a receiver comprising the decoder -- Filed: June 7, 1995 Granted: July 7, 1998 [55]
  • US 6,289,308 -- Encoded wideband digital transmission signal and record carrier recorded with such a signal -- Filed: March 8, 2000 Granted: September 11, 2001 [56]
  • US 5,481,643 -- Transmitter, receiver and record carrier for transmitting/receiving at least a first and a second signal component -- Filed: April 24, 1995 Granted: January 2, 1996 [57]
  • US 5,544,247 -- Transmission and reception of a first and a second main signal component -- Filed: October 25, 1994 Granted: August 6, 1996 [58]
  • US 5,610,985 -- Digital 3-channel transmission of left and right stereo signals and a center signal -- Filed: January 21, 1994 Granted: March 11, 1997 [59]
  • US 5,740,317 -- Process for finding the overall monitoring threshold during a bit-rate-reducing source coding -- Filed: August 30, 1995 Granted: April 14, 1998 [60]
  • US 5,878,080 -- N-channel transmission, compatible with 2-channel transmission and 1-channel transmission -- Filed: February 7, 1997 Granted: March 2, 1999 [61]
  • US 5,960,037 -- Encoding of a plurality of information signals -- Filed: April 9, 1997 Granted: September 28, 1999 [62]
  • US 5,991,715 -- Perceptual audio signal subband coding using value classes for successive scale factor differences -- Filed: August 31, 1995 Granted: November 23, 1999 [63]
  • US 6,023,490 -- Encoding apparatus for encoding a plurality of information signals -- Filed: April 9, 1997 Granted: February 8, 2000 [64]
• Thomson [65]
  • US 4,821,260 Expired [66]
  • US 4,942,607 Expired [67]
  • US 5,214,742 -- Method for transmitting a signal -- Filed: October 1, 1990 Granted: May 25, 1993 [68]
  • US 5,227,990 -- Process for transmitting and receiving a signal -- Filed: January 17, 1992 Granted: July 13, 1993 [69]
  • US 5,384,811 -- Method for the transmission of a signal -- Filed: August 24, 1992 Granted: January 24, 1995 [70]
  • US 5,736,943 -- Method for determining the type of coding to be selected for coding at least two signals -- Filed: May 31, 1996 Granted: April 7, 1998 [71]
  • US 5,455,833 -- Process for the detecting of errors in the transmission of frequency-coded digital signals -- Filed: April 26, 1993 Granted: October 3, 1995 [72]
  • US 5,559,834 -- Method of reducing crosstalk in processing of acoustic or optical signals -- Filed: April 15, 1994 Granted: September 24, 1996 [73]
  • US 5,321,729 -- Method for transmitting a signal -- Filed: April 26, 1993 Granted: June 14, 1994 [74]
  • US 5,706,309 -- Process for transmitting and/or storing digital signals of multiple channels -- Filed: May 2, 1995 Granted: January 6, 1998 [75]
  • US 5,701,346 -- Method of coding a plurality of audio signals -- Filed: September 12, 1996 Granted: December 23, 1997 [76]
  • US 5,742,735 -- Digital adaptive transformation coding method -- Filed: August 25, 1994 Granted: April 21, 1998 [77]
  • US 5,812,672 -- Method for reducing data in the transmission and/or storage of digital signals of several dependent channels -- Filed: December 15, 1994 Granted: September 22, 1998 [78]
  • US 5,579,430 -- Digital encoding process -- Filed: January 26, 1995 Granted: November 26, 1996 [79]
  • US 6,185,539 -- Process of low sampling rate digital encoding of audio signals -- Filed: May 26, 1998 Granted: February 6, 2001 [80]
  • US 6,009,399 -- Method and apparatus for encoding digital signals employing bit allocation using combinations of different threshold models to achieve desired bit rates -- Filed: April 16, 1997 Granted: December 28, 1999 [81]
  • US 5,924,060 -- Digital coding process for transmission or storage of acoustical signals by transforming of scanning values into spectral coefficients -- Filed: March 20, 1997 Granted: July 13, 1999 [82]
  • US 5,703,999 -- Process for reducing data in the transmission and/or storage of digital signals from several interdependent channels -- Filed: November 18, 1996 Granted: December 30, 1997 [83]

I think that this list would need to be cleaned a bit. Some AT&T/Bell/Lucent/Alcatel patents are likely missing, and some of the listed patents do not cover mp3 (ex: US 6,185,539) --Gabriel Bouvigne (talk) 19:25, 11 May 2008 (UTC)

Go for it. I would love a comprehensive list of the patents. Jrincayc (talk) 03:04, 12 May 2008 (UTC)

mp3player

Audio Layer 3, more commonly referred to as MP3, is a digital audio encoding format using a form of lossy data compression.

It is a common audio format for consumer audio storage, as well as a de facto standard encoding for the transfer and playback of music on digital audio players.

MP3 is an audio-specific format that was co-designed by several teams of engineers at Fraunhofer IIS in Erlangen, Germany, AT&T-Bell Labs in Murray Hill, NJ, USA, Thomson-Brandt, and CCETT. It was approved as an ISO/IEC standard in 1991.

The use in MP3 of a lossy compression algorithm is designed to greatly reduce the amount of data required to represent the audio recording and still sound like a faithful reproduction of the original uncompressed audio for most listeners, but is not considered high fidelity audio by audiophiles. An MP3 file that is created using the mid-range bit rate setting of 128 kbit/s will result in a file that is typically about 1/10th the size of the CD file created from the original audio source. An MP3 file can also be constructed at higher or lower bit rates, with higher or lower resulting quality. The compression works by reducing accuracy of certain parts of sound that are deemed beyond the auditory resolution ability of most people. This method is commonly referred to as perceptual coding.[1] It internally provides a representation of sound within a short term time/frequency analysis window, by using psychoacoustic models to discard or reduce precision of components less audible to human hearing, and recording the remaining information in an efficient manner. This is relatively similar to the principles used by JPEG, an image compression format. —Preceding unsigned comment added by 24.236.64.25 (talk) 01:19, 17 September 2008 (UTC)

Bitrate/quality

I'm confused. Several articles on MP3/AAC throughout Wikipedia contradict each other. Here it says that modern MP3 encoders can achieve very good quality at 128 kbps but in 1998 it was equivalent to AAC at 96 kbps and MP2 at 192 kbps. The AAC-HE article states that the quality is excellent at 48 kbps but 64 kbps is lower quality than 128 kbps. So... if 48 kbps is "excellent" then 96 is probably perfect. Therefore, 128 kbps in 1998 was perfect quality??? Maybe the author meant 128 kbps in 1998 was equivalent to 96 kbps MP3 today? That would make much more sense. Also, the average reader has no idea what "192 kbps MP2" would sound like, so best to keep things simple and only compare MP3 quality with MP3 quality from now on instead of introducing numerous different formats? —Preceding unsigned comment added by 70.65.192.118 (talk) 00:05, 7 October 2008 (UTC)

This article is simply wrong. The editors have been drawing incorrect conclusions from the sources they are using. 128 kbps is not excellent, even if it is really VBR averaging 128 kbps, rather than true 128 kbps. --JHP (talk) 20:58, 30 November 2008 (UTC)

128 kbps is not transparent!

The article currently states, "The transparency threshold of MP3 can be estimated to be at about 128 kbit/s with good encoders on typical music as evidenced by its strong performance in the above test." The test being referred to is here. However, as I look through the test results, nowhere do I see the 128kbps compressed audio files being compared to uncompressed WAV files. Transparency does not mean that a compressed audio format is indistinguishable from other compressed formats. It means a compressed audio format is indistinguishable from uncompressed audio formats. Where is the control? The test doesn't demonstrate transparency without uncompressed audio to serve as a control. The last test I saw that compared MP3 side-by-side with uncompressed WAVs found that transparency was achieved at 224 kbps. --JHP (talk) 20:15, 29 November 2008 (UTC)

Let me also point out that in the last graph at the bottom of the page, 128 kbps VBR only scores 4.6 out of 5.0. If we were to assume that 5.0 represents transparency (I don't see a detailed explanation of the test procedures anywhere), then the fact that 128 kbps VBR scores below 5.0 would be proof of non-transparency, rather than proof of transparency. --JHP (talk) 21:06, 29 November 2008 (UTC)

Here's an interesting article on audio quality. It's a shame they didn't test with the LAME codec, though. --JHP (talk) 03:27, 4 December 2008 (UTC)

JHP, if 5/5 is transparent (ergo, perfect quality) then logically 4 and above would be excellent. A 3/5 indicates "adequate" quality as the artifacts are slightly annoying and noticeable without a double-blind test. Anything 4 and above would not be noticed without said test, and a 4.6/5 would not be audible unless the listener concentrated hard on the artifacts. You have a warped opinion of "excellent."

Since I'm too lazy to browse the hydrogenaudio listening tests and edit the article, perhaps you should pay a visit. 128 kbps IS good-excellent quality if the very latest LAME 3.98 encoder is used with VBR. You say you can't notice a difference up until 224 kbps. I regret to inform you that people with your exceptional hearing are in the minority. This article discusses what majority of people perceive as good quality. —Preceding unsigned comment added by 70.65.204.30 (talk) 06:20, 6 December 2008 (UTC)

The issue in question is not "good" or "excellent", but "transparent". Furthermore, how do we know that test subjects define 4/5 as excellent and 3/5 as adequate? Are those the values given to the subjects by definition (i.e. "If excellent, score it a 4. If adequate, score it a 3."), or are you making an assumption of how different test subjects will rate things? Unless told to score it differently, I would naturally consider 5/5 as excellent, 4/5 as good, 3/5 as mediocre, 2/5 as poor, and 1/5 as awful. On that scale, I would say that 4/5 is where adequate begins because I don't want MP3s that are not good.

Again, if the test is just comparing one lossy format against another lossy format, then it is completely invalid with regards to measuring transparency. The only valid way to test for transparency is to compare the lossy format to a format that is guaranteed to be lossless (WAV, AIFF, FLAC, etc.). If that has not been done, then the test should not be used as a reference to make claims of transparency in this Wikipedia article.

It also appears that the test did not control for the quality of the audio equipment. Instead, it looks as if subjects may have taken the tests on their home computers which could have easily had tiny laptop speakers. All of these different issues which can make a test invalid are the reason why I want to read the test procedures. I don't see detailed test procedures being explained on the web site that is used as a reference.

Also, 128 kbps is CBR, not VBR. In the LAME lingo, you can say there is such a thing as 128 kbps ABR, but there is no such thing as 128 kbps VBR. However, unless explicitly explained in this article, I think most casual readers will interpret "128 kbps" to mean CBR because that's what it usually means.

Lastly, I did not say that I can notice a difference below 224 kbps. I said the last test I've seen that compared MP3 (CBR) to lossless audio found that 224 kbps was the point at which absolute transparency was achieved, although 192 kbps was judged excellent but not transparent. I do know that a number of years ago when listening to songs using VBR, I could very easily hear when it switched from 160 kbps down to 128 kbps. I don't know what codec I was using at the time. (FWIW, this past weekend I created my own one-person double-blind test—I was both tester and test subject—and could not hear the difference between "lame --preset extreme -q 0" and lossless on my home audio system. It was not an abx test, but simply a randomized a/b test for my own use.)

I will try to find the test procedures on hydrogen audio. I've been reading through it in the past few days, but so far I've only read an article on MP3, an article on LAME, and a discussion forum. --JHP (talk) 01:41, 9 December 2008 (UTC)

Even the Transparency (data compression) article specifies a higher number than 128 as the MP3 transparency threshold—192 is discussed but not established as the lower limit of transparency. I say take out the reference with its inadequate, no-control test and take out any text that it was supposed to support. Binksternet (talk) 05:33, 9 December 2008 (UTC)

Mr. JHP, since I'm ignorant of Wikipedias quoting mechanism I'll just address each of your paragraphs individually:

(paragraph one)

Because the ABXing software they use has sliders which defines 5 as transparent, 4 as "perceptible, but not annoying", 3 as "slightly annoying", 2 as "annoying" and 1 as "very annoying." Anything 4 and above would not be noticed without a comparison to the raw file, and something like 4.5 definetaly means that it wouldn't be perceptible without direct attention paid to the artifacts. However, many participants on Hydrogenaudio are retards who themselves abuse the rating system and cant do shit without hand-holding.

Also, transparency CAN be reached (I repeat, transparency, not high quality) at 128 kbps for many songs. I have validated this myself many times by conducting my own double-blind tests.

(paragraph two)

I agree 100%. THe tests at hydrogenaudio DO use the raw file as a reference, and I personally never heard of any serious test having an already-compressed file as the reference.

(paragraph three)

Maybe, but high quality headphones are easily affordable, recommended to be utilized for the HA.org tests. I myself use high quality headphones and my evaluations of quality doesnt deviate much from said tests on HA.

(paragraph four)

1998 was 1998, now is now. I dont know how you drew the conclusion that "VBR" is non-existant. However, CBR is still popular so it's possible that readers may interpret it as that. In such a case, clarify it by adding "if VBR mode is used" or simply "128 kbit/s VBR."

(paragraph five)

Yes, CBR. CBR is an outdated concept so I wouldnt be surprised. All modern encoders and devices support VBR. A test conducted half a decade ago using a technique not from this millenium would not be a good reference to use, and should removed from this article.

Yes, 160 kbps should be distinguishable from 128, as 128 kbps commonly smears cymbals, as well as percussion to a lesser degree. --70.65.204.30 (talk) 07:24, 9 December 2008 (UTC)

Just a reminder that comments about personal test results and personal opinions of what makes for a good bit rate don't hold any water in the matter under discussion. The question is about test procedures that may or may not be up to Wikipedia's standard for a reference, and whether to delete the text that such a reference is supposed to support. As for the difference between a score of 4.0, 4.5 and 5.0—if 4.5 had been intended as the line in the sand where audible differences vanish from perceptibility, then the test designer would have dropped the 5.0 completely and made 4.5 the top score. Ridiculous? Yes. That's why I see 5.0 as establishing transparency, not 4.98 or 4.99 or whatever. Gimme five... Binksternet (talk) 08:20, 9 December 2008 (UTC)

Umm, there is no "fact" about what makes a good bitrate. There's a good reason why MP3 is called a "perceptual" coding algorithm. All quality at every bitrate is perceived a.k.a subjective. Which is why subjective (double-blind ABX) testing is best known method for evaluating the quality, and all are "personal opinions" blended into one. Also, where did I say that 4.5/5 is transparent? I said 4+ is not immediately noticeable without a comparison, hence is high quality or "transparent" for all intents and purposes. A 4.0 would be a perceptible flaw but not distracting in any way, a 4.5 would be much harder to notice. I'd give a 4.9 whenever I would notice a difference when concentrating extremely hard but not being able to explain what the difference is, just that there is one.

About providing a reference, browse HA.org's "Listening tests" forum and look for the recent 128 kbps MP3 test. I'd say it definetaly qualifies for a reference, but the fact that any idiot who can cheat the test is allowed to participate personally bothers me. I'd prefer a more official test where strict rules are imposed upon the rating system, and expunging any illegal score like when one participant would rate an obviously low quality sample 4/5 or vice versa. I've seen a couple, such as the 64 kbps AAC vs. WMA done by Microsoft, but most are outdated and don't provide details of what equipment/software/version/settings they used. —Preceding unsigned comment added by 70.65.204.30 (talk) 10:19, 9 December 2008 (UTC)

Transparency means that the audio is indistinguishable from the original. Therefore, in an abx test, a transparent MP3 would get the same score on average as the original (allowing for statistical error). If it gets a lower score, that is proof of non-transparency. Within the range of statistical error, a transparent MP3 would have a 50% chance of scoring better than the original. If an MP3 has any artifacts audible to the human ear, it is not transparent—by definition. Your own statement that "160 kbps should be distinguishable from 128, as 128 kbps commonly smears cymbals, as well as percussion" means that 128 kbps is NOT transparent. How can you claim in one statement that 128 kbps is transparent and in another statement claim that there is a difference in the audio quality between 128 and 160? The fact that they sound different at all is proof of non-transparency.

Also, I never said that VBR is non-existent. I said 128 kbps is not VBR. 128 kbps is a bit rate. VBR is variable bit rate, which means that it changes. Sometimes during a song it may be 128 kbps, other times during the same song it will be a different bit rate. I suggest use of the phrase "128 kbps average variable bit rate" to describe VBR averaging 128 kbps, because it is not 128 kbps all the way through the song.

Also, CBR is not an outdated concept. I believe most online music stores sell downloadable songs at a constant bit rate. Apple uses 128 kbps AAC, except for iTunes Plus which is 256 kbps AAC. Real used to use 192 kbps AAC, but is now selling 256 kbps MP3s. Amazon.com uses 256 kbps MP3s. (I have not actually purchased songs from these retailers since their move to DRM-free songs, so I have not verified these bit rates by playing songs. I'm just going by what they advertise.) --JHP (talk) 02:22, 10 December 2008 (UTC)

How can you claim in one statement that 128 kbps is transparent and in another statement claim that there is a difference in the audio quality between 128 and 160? The fact that they sound different at all is proof of non-transparency. No, I said SOME songs can be transparent at 128 kbps, but for many of them, the cymbals and percussion may slightly be smeared due to MP3's gayass framesize intervals design.

Second paragraph: Let's not get nitpicky here. Text like "128 kbps VBR" is publicly recognized. If the reader wants more detail, he can click on the wikified 'VBR' and learn more about it.

And just because big companies are stupid enough to use outdated technologies doesnt make them less outdated. They aim for compatibility, not quality. Otherwise we would all be using MP4s now instead of MP3s as they are way higher quality. They simply use CBR because not all devices support VBR.--70.65.129.18 (talk) 05:46, 10 December 2008 (UTC)

I said SOME songs can be transparent at 128 kbps, but for many of them, the cymbals and percussion may slightly be smeared... —Are we in agreement then, that 128 kbps is not usually transparent? Can I remove the claim from the article?

Text like "128 kbps VBR" is publicly recognized. —I'll accept that as long as "VBR" is linked to the variable bitrate article.

Otherwise we would all be using MP4s now instead of MP3s —I won't get into a debate over what is meant by "outdated". As far as MP4 (AAC) is concerned, the iPod supports it, but most other devices don't. iTunes Plus downloads are in the MP4 format (.m4a extension), but they're still CBR. I was disappointed to see that Real switched from AAC to MP3.

By the way, Wikipedia does have a special block quote template but I think you can also use the HTML <blockquote> element. --JHP (talk) 09:44, 10 December 2008 (UTC)

I removed the 128 kbps transparency claim. I replaced it with a 192 kbps VBR transparency claim after reading through the hydrogenaudio.org forums. They don't constitute a reliable source, so I didn't use a reference. I'll let other Wikipedians decide whether the bit rate transparency claim should be removed altogether. --JHP (talk) 22:37, 10 December 2008 (UTC)

Dude, what you removed was not a transparency claim. It read "128 kbps was somewhere between annoying and acceptable with older encoders, but modern encoders can provide adequate quality at such a bitrate." "Adequate" and "transparent" are two different things. Change that shit back.

Btw, do you have MSN? Would you be willing to ABX a couple samples and prove you can discern a 128 kbps output from the original?--70.65.209.57 (talk) 06:25, 11 December 2008 (UTC)

I removed that paragraph because I felt the claims made in the paragraph were dubious, but I have put it back for now. However, I have removed the paragraph with the transparency claim altogether. If someone has a reliable source for a transparency claim, they can put it back.

No, I don't have MSN. I didn't realize anyone used MSN anymore. Asking me to do an ABX test is irrelevant because it would violate Wikipedia's rules against original research. However, the ABX test cited in the article proves that 128 kbps average VBR is not transparent. Even this newer test proves non-transparency. If it was transparent, 5.0 would be within the range of the sampling error. --JHP (talk) 09:18, 11 December 2008 (UTC)

I yanked the word 'heavily' and the sentence "While quality around 128 kbit/s was somewhere between annoying and acceptable with older encoders, modern MP3 encoders can provide adequate quality at those bit rates." The words 'annoying' and 'acceptable' and 'adequate' were not proven by the reference and without those words, the sentence was worthless. 'Heavily dependent' was not proven by the reference, either, though the simpler 'dependent' was. Binksternet (talk) 12:34, 11 December 2008 (UTC)

Thank you. Those are the reasons I had deleted the paragraph to begin with. I have removed the "dubious" tag because I don't think the wording is dubious in its current form. --JHP (talk) 00:34, 12 December 2008 (UTC)

I just bought my first MP3s from Amazon.com last night. I need to correct my comment about them using CBR. Although Amazon may list the bit rate for a particular song as 256 kbps, they actually use VBR most of the time. It looks like they are using the equivalent of LAME -V0, which is the highest VBR possible. According to the hydrogenaudio.org forums, that is exactly what they are using most of the time. However, this is all original research. For a reliable source, Amazon lists their MP3 specs here. --JHP (talk) 04:35, 12 December 2008 (UTC)

I doubt the Hydrogen Audio wiki counts as a reliable source, but here it claims that LAME encoding at ~170 kbps VBR and above "will normally produce transparent encoding". I have no idea what empirical tests that claim is based on, but it is still considerably higher than 128 kbps VBR. --JHP (talk) 04:53, 12 December 2008 (UTC)

Hmm... guys, don't you think the formality of references and "original research" is a little ridiculous for proving whether a bitrate of a perceptual codec is annoying or adequate? The proof is in the pudding, and any normal person can easily evaluate the quality himself. Can't we just agree that 128 kbps is OBVIOUSLY not annoying, especially with modern encoders?

I realize a sentence must be supported with a reference, but I personally don't trust tests conducted by big companies as they dont list details and for all we know they could be using an encoder not from this millennium. This is why I like to validate such claims myself with my own unbiased tests. —Preceding unsigned comment added by 70.65.201.108 (talk) 08:26, 12 December 2008 (UTC)

References are not required on Wikipedia, they are just preferred. However, questionable claims are likely to be removed by another editor if they are not backed up by a reliable source. On the other hand, Wikipedia's rule forbidding original research is absolute. Again, the ABX tests and the Hydrogen Audio LAME wiki page both contradict the claim that 128 kbps VBR is transparent. --JHP (talk) 05:30, 13 December 2008 (UTC)

Yes, we've agreed that 128 kbps is usually not transparent. However, if I recall the paragraph correctly, it read something like: "The transparency threshold can be estimated at 128 kbps." Which doesnt imply "128 kbps is transparent." Moreover, claiming that one stating "128 kbps provides adequate quality" constitutes original research is like saying that someone who claims that all elephants have a trunk is also doing such. Here's another reference from a site that's something like Wikipedia, 'cuz it allows any person to evaluate random samples in a double-blind fashion. This is the closest thing to a reliable reference, as they list details and everything, but again, I don't like the samples they use. IMO they dont have a variety of them.--70.65.229.62 (talk) 07:41, 13 December 2008 (UTC)

Saying "The transparency threshold can be estimated at 128 kbps" implies that 128 kbps is transparent. If that's not what it means, then I don't know what the hell it means. Also, I didn't say that "128 kbps provides adequate quality" is original research. I said that your opinion constitutes original research, as does mine. --JHP (talk) 03:37, 16 December 2008 (UTC)

It means exactly what it says: that transparency MAY be estimated at 128 kbps because some audio samples can be transparent at that bitrate, but not all. The original paragraph was: The transparency threshold of MP3 can be estimated to be at about 128 kbit/s with good encoders on typical music as evidenced by its strong performance in the above test. However, some particularly difficult material, or music encoded for the use of people with more sensitive hearing, can require 192 kbit/s or higher.

It clarified itself later by saying that higher bitrates are required for transparency, that 128 kbps is just the "sweet spot" because it can sometimes be transparent, and good/excellent when it cant, and on top of all that: a significantly lower bitrate than ones required for guaranteed transparency.

What part of any of this is misleading or unfair?--70.65.229.62 (talk) 23:05, 16 December 2008 (UTC)

First of all, the references don't even back up that claim. Second, the wording leads readers to think that 128 kbps frequently achieves transparency, when you seem to be arguing here that it will sometimes on rare occasions achieve transparency. Since 128 kbps usually does not achieve transparency, it is no "sweet spot". You really are stretching the truth and presenting weak arguments to make the case that Wikipedia should say to encode at the bit rate that you apparently choose to encode at. I'm done arguing this with you. You haven't made your case. Quite the opposite, the listening tests demonstrate that 128 kbps is not transparent in most cases. --JHP (talk) 02:39, 17 December 2008 (UTC)

I wouldn't say rarely, it does more often than that, though I can't say how much 'cuz I've only done 3 total tests myself where it was transparent once, and I dont know any official test that can confirm this (why would they care?)

A "sweet spot" is a bitrate that's a balance between quality and filesize, and in MP3s case: 128 since it's 50% shorter than 192 for just a minor loss in quality.

Wikipedia should say to encode at the bit rate that you apparently choose to encode at. Actually, I don't use MP3 at all, as it is far outdated IMO, and I think everyone else should also trash it and move on. I got no special interest to promote any particular bitrate of MP3 or MP3 itself, in any way whatsoever. Just like you, I wanna improve the article and make sure newbies are properly informed about the various aspects of MP3. Everything I do on Wikipedia I do with good faith. I'm not perfect but I contribute what I can when I have time.

But, back on topic: why do you think the majority of people encode their music at such bitrate? Because they were brainwashed to perceive such low quality as high? No! Because it sounded perfectly listenable to them and they wouldnt notice a difference without a direct comparison, much like the quality differences between 44 kHz and 22 kHz. But mainly it was to relieve HD space. That was back then. Nowadays, people encode at 128 kbps because the technology improved and gives even better quality than before... and they still save the same space.--70.65.229.62 (talk) 14:21, 17 December 2008 (UTC)

Forgive me for not assuming good faith. Like you, I think MP3 is outdated. I personally favor AAC, but the hardware support on non-Apple devices is lacking. Hopefully that will change as iTunes moves to DRM-free AAC files.

While encoding at 128 kbps was popular a decade ago, I don't believe it is what most people use anymore. Among online music stores, Apple is alone in using 128 kbps, but they don't even use MP3. For years most competing online music stores have been encoding at at least 160 kbps (but didn't use MP3 either). The new norm—looking at iTunes Plus, Rhapsody MP3 Store, and Amazon MP3—appears to be 256 kbps. As for the sweet spot, based on this graph, I'd say the sweet spot for LAME encoding is either -V0 (~240 kbps) or -V3 (~170 kbps). File size declines very rapidly between lossless and 320 kbps with little loss of audio quality, then declines less rapidly from 320 to V0, again with little loss of audio quality. Below V0, file size starts declining at a much slower rate. As for audio quality, it declines very slowly between 320 and V3, then begins to drop off rapidly at bitrates below V3. 128 kbps VBR would be LAME encoding with V5. --JHP (talk) 08:49, 18 December 2008 (UTC)

I didnt say you werent, I was simply describing my position (that I'm not promoting MP3.) The encoding preference of a large entity does not denote the preferences of the public, and isn't a good reference in general. If say Amazon encodes at 256 kbps, what does that prove? That 224 kbps is not transparent? The reason they encode at such high bitrates is to preserve more technical quality -- to make it as close to CD-quality as possible, since bandwidth really isn't an issue like it was a decade ago. 128 kbps MP3s are slowly being replaced by 160/192 on filesharing networks, but a quick search with Limewire clearly shows 128 kbps songs in the majority (and 112/96 for longer songs are also fairly common.) I could take a couple screenshots if this passes off as a reference.--70.65.229.62 (talk) 21:38, 18 December 2008 (UTC)

<--outdent... References are required when there are differences of opinion between editors. I think 128 kbps sounds like ass, making for a strong disagreement between at least two editors on the subject of 128 being "adequate" or not. Adequate for what? 128 is certainly adequate for identifying the song's lyric or artist or beats-per-second. It's adequate for mild background music under noisy conditions. It's adequate for subwoofer bandpass sine wave test tones. It's not adequate in a critical listening situation for a picky listener who knows what kind of artifacts to listen for. This is why we need to find reliable sources in order to back up the text. Binksternet (talk) 15:03, 13 December 2008 (UTC)

128 is adequate for all intents and purposes except professional, where the waveform must be perfect to the bit (they also use a 176 KHZ sampling rate in those studios for extra precision.)

You say 128 sounds like shit, care to prove it? I'll upload a link containing 10-20 seconds samples and you ABX them (see JHP's talk page for details.) And THEN tell me if 128 still sounds like ass.

I'm not trying to get nitpicky or act like a smartass, but I believe your opinion is heavily biased by the low appearance of the number "128" and you suffer from some placebo effect (mini-delusion.) But even if you can discern a difference (you should anyway) and you insist a barely-audible washed-out cymbal artifact bothers you so much, then your opinion is in the far minority. This article, again, discusses what majority of people accept as high quality with great confidence. Besides, why do you think majority of MP3s on filesharing networks are encoded at such a bitrate if the quality sucks as much as you make it out to be?

It's not adequate in a critical listening situation for a picky listener who knows what kind of artifacts to listen for. Well, if you directly concentrate on the barely-audible artifacts and ignore the actual music then I wouldnt wonder. And... if you listen to the music like a normal person then you wont notice artifacts, unless of course the artifacts are so prominent that they're distracting, and that is NOT the case with 128 kbps, but something below like 96, 80, 64 etc. As I said, conduct some actual evaluations yourself and then get back to us.--70.65.239.174 (talk) 12:41, 14 December 2008 (UTC)

You presume quite a lot about the way I listen to music and about the kinds of data compression evaluations I might have taken part in and you end up being wrong on all counts. You also put the number "176 kHz" into the category of "professional" without specifying who, where or for what purpose. Where did this information come from? Hey, I'm a professional audio guy and I don't work in a studio. What number do I use? Every audio engineer has his or her own opinion about what makes for 'transparent' or 'adequate' compression, and rarely do the professionals that I've come across use the same bit rates that satisfy the masses. At any rate (pun intended), this is a talk page for improving the article. My opinion doesn't count, your opinion doesn't count nor would any original research undertaken by me or you. Any discussion that doesn't involve tightening the references isn't a discussion working toward improving the article--what makes the article good is a fair number of on-target references to well-designed tests undertaken by acknowledged experts in the field. Binksternet (talk) 04:53, 15 December 2008 (UTC)

I don't remember, I read somewhere that studios use 4x the sampling rate of 44.1 KHz for accurate engineering and reproduction of the waveform. That wasn't the point. My point was that if you merely want to listen to the music, 128 provides excellent quality. If you plan to engineer that song later, remix it or whatever then you shouldnt be using a lossy codec at all. the way I listen to music and about the kinds of data compression evaluations I might have taken part in so you have ABXed an 128 output with the original?

FYI, I have provided a reference, two of them. Hydrogenaudio's latest 128 kbps listening test and soundexpert.com's 128 kbps page. Both rate around 4.3/5 which logically indicates excellent quality. Now restore that paragraph that says this bitrate provides adequate quality.--70.65.229.62 (talk) 08:29, 15 December 2008 (UTC)

Neither the Soundexpert.com reference nor the Wiki Hydrogenaudio reference characterize 128 kbps stereo mp3 files as sounding adequate or acceptable. I don't think a restoration of those words is supported. Binksternet (talk) 13:43, 15 December 2008 (UTC)

I get it now, anything (even lightly) below transparency is automatically garbage to you. Unfortunately, the public doesn't follow your logic. If the artifacts are not distracting, then the song is perfectly listenable to their ears, thus adequate. And those who cant afford any loss use FLAC.--70.65.229.62 (talk) 14:07, 15 December 2008 (UTC)

I'm tempted to blatently advertise FLAC here. This is the most hilarious discussion ever. Accepted bitrates! On wikipedia! I never seen anything like it, really. Elm-39 (talk) 14:36, 15 December 2008 (UTC)

What is up with the poor performance of LAME 3.98 at 224 kbps in this test? It seems to be doing significantly worse than LAME 3.97b1 with --vbr-new, even though the two algorithms should be essentially the same. (If either were better, it should be the newer one.) --JHP (talk) 02:54, 17 December 2008 (UTC)

All of them (except the last) are rated above 5 which means they're all transparent, hence all sound the same. Though I wonder how the hell their rating system works if it can overlap the maximum. And look at MP2, since when was it ever better (even technical quality) than MP3? Someone should read their PDF and find out how ratings exceed the maximum.--70.65.229.62 (talk) 14:21, 17 December 2008 (UTC)

Since my last post, I read their description of the tests. For audio tests that are near transparency and would normally require golden ears with premium equipment, they use software to audibly magnify any defects. It's how they overcome the problem of not controlling for hearing ability and audio equipment quality that is a flaw in any internet-based audio tests. It seems pretty slick actually. --JHP (talk) 07:22, 20 December 2008 (UTC)

Magnify artifacts, how? You can't without using a lower bitrate. On the other hand, they could probably subtract the 224 kbps output from the original, amplify the resulting difference and mix to original, but clearly this is pointless, especially if MP2 was rated better than MP3.--70.65.229.62 (talk) 22:46, 21 December 2008 (UTC)