This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.ComputingWikipedia:WikiProject ComputingTemplate:WikiProject ComputingComputing articles
This article is within the scope of WikiProject Microsoft, a collaborative effort to improve the coverage of articles relating to Microsoft on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.MicrosoftWikipedia:WikiProject MicrosoftTemplate:WikiProject MicrosoftMicrosoft articles
This article is within the scope of WikiProject Microsoft Windows, a collaborative effort to improve the coverage of Microsoft Windows on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.Microsoft WindowsWikipedia:WikiProject Microsoft WindowsTemplate:WikiProject Microsoft WindowsMicrosoft Windows articles
In the bullet point quoted below, the text "a roughly equivalent quality than MP3" seems to contain a grammatical error, or is at least very confusing.
"At mid-low bitrates (64 kb/s or more, less than 128 kb/s), latest private tests (80 kb/s (2005/07), 96 kb/s (2005/08)) show that WMA has a lower quality than the lossy audio codecs AAC (HE and LC) and Vorbis, a roughly equivalent quality than MP3, and a better quality than MPC."
This screenshot is a bit weird. VLC can't even play a multichannel WMPro file (or was this added recently?). Anyway, it's inappropriate. Add one of Windows Media Player, the native player for these codecs! —Preceding unsigned comment added by 83.87.147.223 (talk) 11:27, 6 July 2010 (UTC)[reply]
I've added links to the largest, double blind and independent listening tests I know of twice now, and both times they've been backed off and replaced with nonsense about how the difference has not been determined. Anyone who has looked at the evidence can see this is not the case.
Could I get some clarification how exactly citing the conclusions from double blind tests is biased? Its not as if the people evaluating the formats even knew what formats they were, so how could they be biased? As far as I am aware, these are the two largest, and most comprehensive tests of the formats, and they have been reviewed and accepted by developers working on a number of open and closed source audio codecs as well as virtually the entire audio testing community and most of the less insane audiophile webistes. Calling the results biased is ridiculous. —The preceding unsigned comment was added by 150.135.157.10 (talk • contribs) .
The link is fine, but the prose is grossly unencyclopedic. It has all the appearance of trying to add a non-netural POV to the article, and you will get a lot of push-back on that. Warrens18:24, 17 April 2006 (UTC)[reply]
The linked article wasn't nearly as cut-and-dried as the most recent revision of the Wikipedia article made it seem. I've kept the link, but added some clarifying text in an effort to reach an NPOV medium. Raider Duck18:42, 17 April 2006 (UTC)[reply]
At least from my personal experience, Windows Media Audio has horrible high ends. Any track with a lot of cymbals sound absolutely horrible, and very liquidy. This may have changed since I last used it, so, I don't want to write about it unless this can be verified by other people.--68.196.38.1318:32, 9 October 2005 (UTC)[reply]
I've found this is a problem with all popular lossy audio formats, including AAC and MP3. I find that I have a acute sense of hearing in the really high frequencies that most people don't hear though (dog whistles, old cathode ray tubes, and damaged capacitors really bother me) so it may be that the engineers that created these formats didn't take into account that the range of hearing of individuals varies widely.
Due to the nature of the compression, Windows Media Audio and MP3 files at 128 kilobits persecond have virtually no difference in sound quality. However, the difference between WMA at 64 Kbits and MP3 at 128 Kbits is also nearly indistinguishable. The quality comparison of the two music compression formats can be found on the microsoft website [1]
If you look at the files, you will see that this is not a fair comparison. The 128kb/s samples are fine (and both sound virtually the same.) However, while the 64kb/s samples do have the same bitrate, they have different sample rates. The wmas are 44100Hz and the mp3s are 22050Hz. With the 32 and 20kb/s samples, it is 22050Hz wma vs. 11025Hz mp3. The cuts in the higher frequencies are very significant in the given samples which are fairly high guitar (type) music.
I am removing the section discussing Janus (v10) DRM because the DRM technology is not a part of the codec technology, just like Apple's FairPlay technology is completely unrelated to the AAC codec technology.
The goal of Wikipedia project is to make an encyclopedia, not a well of rumors and slanders. I propose to remove Sound quality section as long as there is no claim originating from a reliable professional source such as an independent sound studio or university research paper.
--Ondrejsv15:49, 20 May 2006 (UTC)[reply]
Are you the same one who edited this section a few hours ago? I don't see anything wrong with the hydrogenaudio listening tests. In fact, both Apple Computer and Nero AG have paid close attention to these tests and worked to improve their codecs as a result. Why do you consider this "rumors and slanders"? --Mcoder17:33, 20 May 2006 (UTC)[reply]
Well, the problem is how these "tests" were done and by who. They are not prepared by a professional who knows which parameters to test and can carefully prepare sound samples. Also, to deduce results, you must also know something about statistics and reliability levels. Besides this, these tests are only subjective "good/bad" checks, no analysis or real research has been done. And frankly, you absolutely don't know the people who ranked the samples. --Ondrejsv14:52, 29 May 2006 (UTC)[reply]
Regarding the test process for the group tests referenced: the double-blind listening comparison program used in these tests (abchr-java) has the capability of performing both time-synchronization and level-matching, which are required to perform proper listening tests. See http://ff123.net/abchr/abchr.html for the original Windows version of the abc/hr program. The codec parameters are either solicited directly from the codec developer or chosen by consensus from within the community interested in the comparison during a long period of pre-test discussion. It is true that the people who agree to test the samples are an unknown quantity -- a self-selected group of (I'm assuming) audio enthusiasts using various equipment and listening environments. But this may actually be more applicable to real life than a group of trained listeners sitting in a standardized environment. Results tampering is minimized by encrypting the test results (an encryption key is provided after the results have been analyzed). The statistics are not difficult (blocked multiple comparison). See http://ff123.net/friedman/stats.html for the web version of what is used for the analysis, along with source code for the truly interested, and yes, confidence levels are provided. All raw test data is made available so that independent analysis can be performed, if desired. In the end what matters is not whether the tests were performed professionally or in a educational research setting, but whether they are meaningful. I don't think it's fair at all to imply that these test results are no better than "rumors and slanders." Ff12306:53, 5 April 2007 (UTC)[reply]
There are no sources for them. I'd like to see some sources for statements like "WMA Standard is the second most popular digital audio compression format after MP3"
"More than 500 models of consumer electronic device, from car stereos to home theater systems, now support Windows Media Audio (WMA) or WMV, including products from Matsushita/Panasonic, Pioneer, and Thomson/RCA—a number that has increased 150% since Jan. 2003. In addition, more than 4 million WMA-capable portable music players from companies such as Creative Labs and iRiver have shipped since 1999, making it the second most popular format for these devices after MP3. (Three and a half million portable MP3 players shipped in 2003 alone, according to Jupiter Research.)"
It was the second most popular format for a long time, and may still be, but its hard to nail it down. Do we count by number of users? Number of tracks? Hardware licenses sold? IMO simply saying something like "WMA Standard and AAC are the most popular formats after MP3" would be better. 152.3.198.22102:52, 30 July 2007 (UTC)[reply]
It should be judged by the number of files. Definitely not number of standalone players, as many people keep/play their audio files on computers or phones. That a device is WMA-capable doesn't mean people actually use it to play WMAs. Also, these data are from 2003. I would suspect Apple has taken over in the meantime, especially since AAC is a standard now so everything that is MPEG-4 compliant plays AAC.--87.162.34.1811:21, 18 September 2007 (UTC)[reply]
AAC was standardized in the late 90s, so I don't see what 2003 has to do with the standardization process. Anyway, lots of MPEG4 compliant software and hardware does not support AAC. While AAC is a part of the MPEG4 standard, so is MP3, MIDI and VQF. Generally devices pick the codecs they need, and many of them chose MP3 over AAC (since there is a separate fee for each and its cheaper to use fewer). Most set top Divx mpeg4 players for instance chose mp3 over aac for cost reasons. —Preceding unsigned comment added by 152.3.198.221 (talk) 03:52, 18 October 2007 (UTC)[reply]
If you want quality over *absolutely anything else*, WMA Lossless should be chosen. But it needs about 350-400 MB per 80 minutes CD. For 16 bit 2 channel audio (regular CDs), Pro is probably overkill, standard CBR or VBR will suffice. If you are using CBR, a bit rate of 192 Kbps will be enough for most uses. If you have a very high end sound system, you might want to rip at 256 Kbps. I personally prefer using VBR, though. It might not reduce file size (in some cases it might result in little bigger file) but they tend to have somewhat better quality. I rip at the 135-215 Kbps setting (in WMP 11). That works for me.
But since everyone's listening tastes might be different, try encoding 4-5 CDs with the different settings and settle for the one you feel gives the best balance of quality and size, per your needs.--soumtalk06:33, 3 August 2007 (UTC)[reply]
I am not going to use the lossless format - I consider that overkill. I just got the impression that the Pro format was a vastly superior next generation format. Since it apparently does not take up much more space, it seems like the best option. Of course, there is the problem of lack of hardware support, but I expect to be playing from the PC almost exclusively. Also, there should be good options for converting Pro to for example VBR - right?
I just want to know whether there will be any difference in sound quality at all going from Pro to VBR when ripping the music from CD's - if not, VBR with its smaller file size and better support it is. --HenrikSH16:16, 3 August 2007 (UTC)[reply]
The VBR "codec" you see in WMP11 is really WMA Standard using VBR instead of CBR. The Pro codec has better quality than Standard so you need a higher bit-rate for Standard to get the equivalent quality of Pro.
Since you said you prefer quality over size but not lossless. Try ripping in Lossless and then batch encode to Pro at 256kbps using Windows Media Encoder 9. This is the highest bit-rate setting for stereo, 44.1kHz, 16-bit (CD spec) for this codec.
By the way, if you decide to take the Pro path using Windows Media Encoder, don't forget to set 2-pass CBR to squeeze out the last "juice" out of your bit-rate. Nicholas202001:43, 6 August 2007 (UTC)[reply]
Trying to get a better rating for this article. What do you think should be done to get anything like FA/A/GA for something as invisible (lack of images) as codecs? Nicholas202001:26, 9 August 2007 (UTC)[reply]
Nice work on the article. What I find lacking is information on the actual algorithm (is there any published study or patent app out there?) And the stream structure may not be accessible to non-techy users. You might consider some jargon busting (like definining what are superframes or bit reservoirs when you mention them). --soumtalk07:11, 9 August 2007 (UTC)[reply]
Thanks soum for the feedback! I unfortuneately wasn't the one who worked on the algorithm part so we'll need to get some research data on that. There are obviously some reference "holes" in the article which we need to work on.
The feedback I've got so far is that if we want to get the article to FA we'll need a Histry section, and some images, which is a little difficult because sound isn't the most visual thing for an article. Anyway hope everyone can contribute! : ) Nicholas202008:37, 11 August 2007 (UTC)[reply]
This is a great article; it informed me without overwhelming me in technical jargon. Although I don't know a lot about this topic, I would suggest that you expand (if possible) the parts on players, encoders, and digital rights management. Lando523:01, 29 August 2007 (UTC)[reply]
No, no, and no. The Discussion Page is also not the place to ask such questions. May I suggest you use Bittorrent or something? And don't get sued by the RIAA. - 68.228.41.7007:58, 14 November 2007 (UTC)[reply]
WTF? I did not mean music files from P2P networks. I just meant if WMA files can be uploaded. WMA is as popular as MP3, and Cortado is EXTREMELY buggy, which pisses me off. --208.138.31.76 (talk) 20:44, 21 November 2007 (UTC)[reply]
I think the entire page needs a rewrite. From the here-and-there POV comments to the endless section on a stupid comment Microsoft made a long time ago to promote WMA, this page is slipped to crap. Needs a rewrite. - 68.228.41.7008:00, 14 November 2007 (UTC)[reply]
Unlike the other articles on Wikipedia on audio/media formats, this article does not contain any information on how Microsoft licenses WMA. Can someone add it please? I could not find accurate information on Microsoft's website. —Preceding unsigned comment added by 221.128.181.45 (talk) 06:56, 31 December 2007 (UTC)[reply]
Inthe last revision I edited, I found duplicate named references, i.e. references sharing the same name, but not having the same content. Please check them, as I am not able to fix them automatically :)
"streamingmedia%com Windows Media Best Practices" :
{{cite web |author=Smith, Tony |title=Best Practices for Windows Media Encoding |date=[[2007-02-21]] |url=http://www.streamingmedia.com/article.asp?id=9510&page=2&c=4 |accessdate=2007-08-16 }}
<ref name="edn">{{cite web|url=http://www.edn.com/index.asp?layout=article&articleid=CA6391444&industryid=22043|title=Voices: Microsoft's Amir Majidimehr: a window to the world of digital media - 11-23-2006 - EDN}}
Does the term "WMA Pro" refer to WMA Professional, or WMA Protected? --Atlantima (talk) 13:47, 13 August 2008 (UTC)
With WMA Protected you maybe mean WMA DRM (Digital Right Management) protected files, WMA Pro uses another encoding —Preceding unsigned comment added by 83.64.248.68 (talk) 14:31, 26 August 2010 (UTC)[reply]
i've added the fact tag to the "Sound Quality Claims" part of the Criticism paragraph - i've never seen anyone claim that 128kbps MP3 is commonly thought of as transparent. until someone can actually find a source for that bit, it's just baseless speculation. Onesecondglance (talk) 21:16, 26 August 2008 (UTC)[reply]
If my memory serves me, in the early days, 128 kbps MP3 was commonly called “CD quality” in programs that allowed you to create MP3s. I think it had more to do with 128 kbps being the minimum bitrate at which early encoders would allow you to use 44.1 kHz (the sampling rate specified in the Red Book audio standard). Users misunderstood it as meaning that the end result would sound as good as a CD in every way. — NRen2k5(TALK), 23:01, 14 May 2009 (UTC)[reply]
In a 1999 study funded by Microsoft, NSTL found that listeners preferred WMA at 64 kbit/s to MP3 at 128 kbit/s (as encoded by MusicMatch Jukebox).[55] However, a September 2003 public listening test conducted by Roberto Amorim found that listeners preferred 128 kbit/s MP3 to 64 kbit/s WMA audio with greater than 99% confidence.
Microsoft held that study because they wanted to say that listeners preferred a soundfile encoded at half the quality to sound just as good as good as the standard 128 kbps of an MP3, but it proves nothing if you do the reverse, so maybe the Wikipedia article writer just flubbed this part, the report that the second study reversed this to try to prove that a 128 kbps file sounds better than a 64 kilobyte per second one does (of course it does). —Preceding unsigned comment added by 76.122.119.33 (talk) 10:51, 9 November 2008 (UTC)[reply]
You can figure it out even from reading your own comment:
Microsoft held that study because they wanted to say that listeners preferred a soundfile encoded at half the quality to sound just as good as good as the standard 128 kbps of an MP3.
Microsoft’s study wasn’t objective. It’s goal was just to make WMA seem superior, not an objective comparison. Mr. Amorim’s independent, objective study disproves Microsoft’s claim. And that’s the actual point of including Mr. Amorim’s study in the article: to show that Microsoft’s claim was false. — NRen2k5(TALK), 23:17, 14 May 2009 (UTC)[reply]
In the section Windows Media Audio Lossless the following FALSE claim is written:"When decompressed, the audio signal is an exact replica of the original." Only if the "original" signal is digital can there be "an exact" replica, and even then you'd need to use world class clocks and electronics to obtain it, neither of which is going to happen. No ADC conversion is "exact", just as no DAC is exact going the other way. I think this line should be either deleted, or modified to indicate that the replica can result in a signal indistinguishable from the original (although this claim should have pretty strong references to back it up: average human ear? audiophile? electronics (sampling at, say, 1000 Terahertz)?) (and I won't get into the fact that a finite process can NOT compute an infinite series (such as a cosine))173.189.72.93 (talk) 20:15, 14 November 2014 (UTC)[reply]
Frankly that part is a mess. AFAICT there is only one loseless codec not two as the section may lead people to believe. That said, I think this is a bit of a silly debate. The wording may not be the best, but it's clearly referring to the original digital audio, not some other theoretical audio signal. Probably the wording in the section above is a little better "The result is a bit-for-bit duplicate of the original audio file" although also not quite right. (It's only bit-for-bit when decoded. Also file should be used with care since the actual decoding output could be in a different format from the audio file, probably audio stream would be a better choice.) BTW, your 'is digital' and 'world class clocks and electronics' is a little weird. If I generate a 8kHz tone with some software, I don't need 'world class clocks and electronics' to obtain it. I just need to keep the output of my software and only ever encode it losslessly. The same as if I use some midi rendering software to render my midis. (Probably the samples for the midi rendering were originally analog, but the only logical definition of the original audio signal of my rendered midi is the output of the software I used, no matter what form the samples were in. And of course you could use samples that were only ever digital or use FM to render your midi or whatever.) You seem to be making the assumption an audio signal must generated physically to be an audio signal. BTW I choose midi rendering because it's an example of a complicated audio signal which can easily be considered music, which is so far any original audio that it seems clear it's silly to refer to some original physical audio signal. But in reality a lot of music nowadays has similarities with different parts being recorded in different sessions, some parts maybe being generated mostly digitally, auto tune and other processing etc; such that the final output audio signal is clearly and intentionally different from any one physical audio signal you can identify. Nil Einne (talk) 18:22, 14 January 2019 (UTC)[reply]
Fixed —Codename Lisa (talk) 15:31, 5 April 2016 (UTC)[reply]
