Wikipedia talk:Music samples

From Wikipedia, the free encyclopedia

Contents 1 Samples quality 2 Question 3 Other discussions 4 This endorses music samples too much. 5 Inline inclusion 6 New additions 7 Some thoughts 8 Bad examples 9 Thoughts on "Quality" 10 Good Examples 11 Straw poll 11.1 Policy 11.2 Guideline 11.3 Oppose 12 Template example in Guideline 13 "Relevant samples" 14 Special:Unused media 15 CSD criteria 16 Classical Music Samples 17 Proposed CSD criterion 18 It's a Guideline 19 Lack of clarity 20 Backwards samples 21 Exceding fair use in sample length or in number of samples for album 22 Audio from television episodes 23 Where does 30 seconds come from?

[edit] Samples quality

What, precisely, does it mean that the samples must be of reduced quality? The {{Music sample}} template says that they qualify as fair use when, among other things, they are of an inferior quality to the original recording. But isn't it always true when we use lossy compression format? Jogers (talk) 12:02, 13 June 2006 (UTC)

Thank you for opening the discussion page. I noticed that samples uploaded in Wikipedia are of a high quality. You can see it by their size: Cool (song) has a sample of 500 Kb and Hollaback Girl has one of 1300 Kb. Plus, I made a comparision between samples in Wikipedia and in Allmusic.com. Allmusic.com's samples are of a lot less quality. One way of resolving this problem is to suggest a clear bitrate value, although I don't know much how ogg files work. CG 13:07, 13 June 2006 (UTC)

The fair use sound samples quality is an important issue that has to be resolved as soon as possible. I've checked some samples with a frequency analysis tool in Adobe Audition and they all seemed to be of pretty high quality (the sound quality is not necessarily defined by the bitrate alone). Does All Music Guide use the samples of lower quality because of copyright issues or just because of the storage limits? I can't tell because I don't know much about the law and the fair use doctrine. I would say that lossy compressed files are always of an inferior quality to the original recording but sometimes the difference may be extremaly insignificant and hard to distinguish. I don't know whether using such samples on Wikipedia is a copyright infringement or not. Jogers (talk) 16:36, 13 June 2006 (UTC)

I think that AMG uses low quality samples because of copyrights issue, I doubt that there is a storage limit. As for this issue, I won't be able to help, therefore, we will need users who knows enough about Ogg format and sound quality in one hand, and in the other hand about fair use and copyrights. We might as well put a notice somwhere. CG 17:03, 13 June 2006 (UTC)

The fact of the matter is that Ogg Vorbis does a fantastic job of compression without distortion in sound files, and that using the lowest Vorbis quality setting (level 0, roughly equivalent to 64kbps) is more than sufficient for song samples. Take the song samples I provided for the Rammstein article: Sample 1, Sample 2, and Sample 3. They sound pretty good, and the compression is really only noticeable if you directly compare them to the originals. They give the user an idea of what the song sounds like and illustrate the musical techniques, while satisfying our reduced quality requirement. So with that in mind I've added a suggested quality setting. ~MDD4696 13:50, 14 June 2006 (UTC)

The issue here is not finding the best audio compression format, but to set the bitrate that will produce clear low quality sound. We don't want to get sued because we the compression format we're using produce apparent high quality. Anyway I compared the samples you gave me with the ones in AMG. I noticed that AMG's are of a lower quality (go try it). And if you say that it's the minimum quality we could produce, then we'll have to find something else. And by the way, I'm not sure, but I tink that samples provided by AMG have a bitrate of 20 Kbs. CG 14:45, 14 June 2006 (UTC)

What I was trying to say is that we can use a quality setting of 0 with the Vorbis codec, and it will result in a lower quality sound file that is still clear enough to use. It's ok that we're using higher bitrate sound files than AMG--if you compare my samples to the originals there is a significant difference in quality. ~MDD4696 22:31, 14 June 2006 (UTC)

Are samples to be produced at a clearly low quality, or a clearly reduced quality. I'm considering that there may be music included (perhaps from the early 20th century) that would already be of relatively low recording quality. Ultimaga | Talk 22:00, 19 June 2006 (UTC)

[edit] Question

does this policy apply only to non-licensed (e.g, fair use) music samples, or all music samples? Raul654 12:40, 13 June 2006 (UTC)

samples of non-copyrighted songs face a lot less restrictions, but we may also add that Wikipedia is not a collection of photographs or media files. CG 13:07, 13 June 2006 (UTC)

I am thinking specifically of how this proposal would affect sections like Mozart#Media, Ludwig_van_Beethoven#Media, etc. Those articles are greatly enhanced by having full length songs in them (and said songs are not easy to find). Also, wikipedia:sound/list Raul654 13:18, 13 June 2006 (UTC)

Raul654 has a point. The guideline should reflect the usage of copyleft samples too. They don't need to be at most 30 seconds long and they use different licensing information. Jogers (talk) 16:45, 13 June 2006 (UTC)

We could always mention that. The only problem, a minor one, I could think of is because these songs are free, uploaders might upload tons of unused songs. This guideline should set a limit (see [[Wikipedia:What Wikipedia is not#Wikipedia is not a mirror or a repository of links, images, or media files. CG 17:03, 13 June 2006 (UTC)

[edit] Other discussions

I forgot to display important matters that should be discussed. Music samples are widely used in Wikipedia and violating copyrights is also easy.

• We should agree that song samples are necessary and every song (at least the ones with an article) deserves a sample in Wikipedia.
• I'm concerned about problems that will occur when a user disagrees with the section of the song the sample displays and upload another one of the same song.
• Requesting for concerned users to create a special page called Special:Unused media (much like Special:Unused files, which will help us spot unused or duplicate samples.
• Should we remove every high quality samples and replace them by low quality? Should there be a maximum bitrate?

Your input is appreciated. CG 13:04, 13 June 2006 (UTC)

Not every song, not even those with an article, needs a sample. The test ought to be "does the sample help the reader understand the article?" If the song is noteworthy for musical technique, it absolutely needs a sample. If the song is noteworthy for having been a chart-topper, it doesn't need a sample. If an article about an artist/band refernces musical style, and example of that style is appropriate. GRBerry 14:36, 16 June 2006 (UTC)

I agree with GRBerry. We should only use samples where they actually enhance our understanding of the song. ~MDD4696 17:12, 16 June 2006 (UTC)

I don't follow. Doesn't a sample always help the reader understand the article about a song? I don't mean that every mentioned song should have a sample but having a sample in every article about a song sounds fair. Jogers (talk) 09:45, 17 June 2006 (UTC)

Suppose we're talking about music by Madonna. In general, if you have heard any of her music, you know what it generally sounds like. Suppose we have three articles on individual songs by her: one is very different from her general music, one was a huge hit, and one is not. In my opinion, the one that is unusual for Madonna and the hit should have song samples; the reader does not need a sample for the generic song because it sounds like all of her other music. Users can get a general idea of her style on the Madonna page.

This brings up a corollary: there are a lot of articles about individual songs that don't really deserve their own article. We could use song samples as a guideline there--only create a page about the song if you could justify using a song sample for it. ~MDD4696 15:08, 19 June 2006 (UTC)

And if it seems like I favor restricting non-free content as much as possible, you would be correct. We're building Wikipedia, the free encyclopedia here. ~MDD4696 16:48, 19 June 2006 (UTC)

Can we at least agree that every article about a musician or a band needs a sample? Jogers (talk) 08:53, 21 June 2006 (UTC)

[edit] This endorses music samples too much.

Keep in mind that music samples are almost always copyrighted and unlicensed (and everything tagged with {{Music sample}} is). "There's no limit of how many samples you could use in one article, but you have to put in mind that music samples serve as tools for a better understanding of the article, so insert only relevant samples" says that the only reason you shouldn't have a particular music sample in a particular place is because it may not be relevant. Remember that they're fair use, so they should not be overused. Wikipedia:Fair use#Policy says "The amount of copyrighted work used should be as little as possible. Low-resolution images should be used instead of high-resolution images (especially images that are so high-resolution that they could be used for piracy). Do not use multiple images or media clips if one will serve the purpose adequately." --Rory096 03:43, 14 June 2006 (UTC)

The phrase of the Fair use policy you mentioned could be interpreted in many ways: It could be seen as it imposes a limit or the contrary. Anyway, I don't think it's the main problem because many samples violates the low-resolution requirement and are of a high quality (see the discussion above). I created this proposal to set clear limits as to samples quality and durations. As for the relevancy of samples, I prefer that there always be discussion concerning which sample should be uploaded and which secton of the song should be represented, rather than uploading any song a user like. CG 08:52, 14 June 2006 (UTC)

[edit] Inline inclusion

Isn't the {{Audio}} template meant to be used for stuff like pronunciation etc? Jogers (talk) 09:43, 14 June 2006 (UTC)

I don't like it, but look at Genesis (band), it's used a lot. CG 10:36, 14 June 2006 (UTC)

Genesis (band) uses it a lot wrongly--many of those samples should be removed. I can see how inline links would be useful, but I think an audio sample box is the best solution. It makes it easier to find the samples, and it makes having too many samples on a page cumbersome. ~MDD4696 13:59, 14 June 2006 (UTC)

[edit] New additions

Thank you for the rewriting of the intro, it's better now. About the 10% figure you gave, is it applicable? most songs have a lenght of 3-4 minutes, that means 180-240 seconds. Their 10% samples would be 18-24 seconds, which is always less than 30s. CG 14:53, 14 June 2006 (UTC)

Well, yes. See http://www.utsystem.edu/OGC/IntellectualProperty/ccmcguid.htm. ~MDD4696 22:24, 14 June 2006 (UTC)

I know the RIAA considers <7 seconds to be de minimis. Raul654 05:01, 15 June 2006 (UTC)

The more I read, the more I see that copyrights issues are complicated. As for the link you gave me, it states that it should be up to 10%, but in no event more than 30 seconds. Anyway, this proposal will need more experiences users. CG 07:32, 15 June 2006 (UTC)

[edit] Some thoughts

I'm very happy to see this guideline - although publishing music samples is, to some extent, playing with fire, I believe it would immensely add to the value of the articles they augment. I'd be happy with almost anything as long as we can all settle on it and get started using it. I am thankful once more than En allows fair use.

At first I thought the guideline was a little too strict in establishing the exact number of seconds, but I think this kind of hard rule will make people feel more comfortable uploading relevant samples. On the other hand, it might also provide an excuse for uploading irrelevant samples - I would consider adding a rule that a sample should only be uploaded if it can immediately be used right next to some directly relevant article text. This strengthens fair use and cuts gratuitous uploads. Deco 09:12, 15 June 2006 (UTC)

I'm glad that many users are interested in this important proposal. I'm for your suggestion: Any sample that is not used can be deleted immediatly or Never upload a sample unless you want to use it immediatly in an article. We might as well make a WP:CSD criteria. CG 10:57, 15 June 2006 (UTC)

[edit] Bad examples

Kylie Minogue's "I Should Be So Lucky" is 3 minutes 25 seconds in length. Her version of "The Locomotion" is 3 minutes 11 seconds. "Silent Sun" was a very short song, only 2 minutes and 15 seconds. 30 seconds is not 10%. Someone should takes the shears to these and clean out the old versions. Deco 09:27, 15 June 2006 (UTC)

Also, these sound pretty good to me - not as crisp as a 128 kbps, but no audible clipping or noise artifacts. Is this really as far as you can "turn down" the quality on OGG? Can you supply a lower bitrate argument or something? I would not advise the deliberate destruction of quality for purposes of fair use - that's just silly. Deco 09:32, 15 June 2006 (UTC)

I don't know much about the copyrights rules, but it seems that 30s is OK, but it's a limit that should never be crossed. About quality, the less the quality is, the safer we are. I think that all samples in Wikipedia, even the one proposed as very low quality, are still high quality to my ears. I recommend you to create an account (it's free) in Allmusic.com and compare some samples, you see they use much less quality but it stills audible. CG 10:57, 15 June 2006 (UTC)

All Music Guide uses samples of very low quality. The samples at Rolling Stone are much better. I think we should use samples of as high quality as legally possible. Jogers (talk) 11:07, 15 June 2006 (UTC)

But the proposal says 10%... Deco 13:48, 15 June 2006 (UTC)

Deco, the lowest possible quality setting in Vorbis is -1. The clips you referenced were compressed at a quality level between 4 and 6 (depending on the clip). We should all use a quality setting of 0--at a quality level of 0, the audio is significantly compressed, and somewhat degraded, but still retains sufficient fidelity for encyclopedic analysis. Secondly, as I said above, the length rule is no more than 10% of a song, up to 30 seconds. This means that for many modern songs, samples should not be 30 seconds long, because it would be more than 10% of the song. ~MDD4696 14:17, 15 June 2006 (UTC)

Right, what I'm saying is: this page cites them as an example, but they don't comply. We should use a sample that does comply. Deco 07:39, 17 June 2006 (UTC)

We definitely need better examples. According to the current guideline these are too long and of too high quality. Jogers (talk) 09:51, 17 June 2006 (UTC)

[edit] Thoughts on "Quality"

I'm generally in favor of the proposal, but there seems to be some confusion about the meaning of 'low quality' samples. By definition, all lossy compression formats are lower quality than the source. At high enough bit-rates, people begin to loose the ability to tell the difference, depending on their equipment and their ears. I'd caution us against creating music samples that are too low in quality. I frequently find myself looking at an image and wishing I could get a closer look, or listening to a song and wishing I could hear it a little better. The benefit of a digital medium like this is that we can actually have higher resolution media if we want it.

The legal requirement of 'lower quality' should be met with as high a resolution as we can justify. Ogg Quality 0 (approximately 64kbs) is decent, but I feel we could legally recommend a higher quality. Q=5 is too high, Vorbis files experience a jump in file size at 5, due to a change in how joint stereo information is processed. However, 3 or 4 would be noticably degraded in quality (certainly not a replacement for the original, which is the point), yet clearer than the FM radio-esque Q=0 files. Wikipedia isn't meant to be a repository for media, but for the small amount of media we justify keeping around, the quality should be as high as technically and legally possible. Phidauex 21:25, 17 June 2006 (UTC)

Additional notes: We've been discussing both Ogg Quality Settings and Bitrate, which are not directly correlated. People are used to encoding MP3s with a fixed bitrate, meaning a 96kbs file sounds worse than a 128kbs file. However, in Ogg Vorbis, you do not specify a bitrate, you specify a "Q" value, normally between 0 and 10. The format is adaptive and variable, and so two songs of identical length, recorded at the same Q setting, may result in different average bitrates. If OGG Vorbis is the preferred format (and I think it should be), then we should NOT specify a bitrate, but instead specify a Q (quality) setting. Phidauex 21:32, 17 June 2006 (UTC)

One thing we could do, although this is a bit strange, is to upload all the OGGs at high quality, but only allow them to be downloaded (whether from the article or from the media page) in a wiki-wide quality setting. The software resamples on the fly. Maybe too processor intensive. I think realistically the best thing we can do in the short term is just settle on something we can all agree on. If only we had a lawyer handy with legal experience in fair use. Deco 22:17, 17 June 2006 (UTC)

Do my comments above fall on deaf ears? A quality setting of 0 is sufficient for most all of our purposes... therefore, in most cases, we would have a hard time justifying high quality samples. Deco, you should read up on UMG v. MP3.com. The Wikipedia article doesn't specifically mention it, but one of the points emphasized by UMG was that MP3.com had no authorization for the original copies of the songs stored on the servers. It seems to me that even if Wikipedia served only degraded quality versions of the high quality files, there still might be a problem. ~MDD4696 02:12, 18 June 2006 (UTC)

I think you're right. Better not to play chicken with fair use - it's such a big fuzzy line that there's no telling what side we'll fall on if we dance around the high end. Let's just select something that sounds of comparable quality to what is heard on most other sites. As far as I know OGG quality 0 seems to fulfill this. Deco 07:00, 18 June 2006 (UTC)

Agreed... While the technical feature of OGG files to be downgraded upon streaming (which is done by stripping bits, not 'resampling' so it requires almost no server overhead) is there, the storage of the high quality files would probably be a problem. Its clear that copyright law is not keeping up with our changing world. Most of the fair-use guidelines we are discussing aren't law persay, but are proposed guidelines. Its up to individual courts to make up their interpretations as they go, which may or may not fall on the same lines as the proposed guidelines. Even holding to the proposed guidelines might not be sufficient. I suppose an OGG Q=0 standard would be the safest.

An excellent resource for non-laywers wanting to know more about fair use: http://fairuse.stanford.edu/

Phidauex 17:33, 18 June 2006 (UTC)

Cornell and Stanford have such great law reference sites... I love it :). ~MDD4696 18:30, 18 June 2006 (UTC)

Personally I think some of these guidelines may be a bit overly conservative. Keep in mind that Wikipedia would typically be allowed broader fair use applications than a commercial website. As the Copyright Act of 1976 states:
"In determining whether the use made of a work in any particular case is a fair use the factors to be considered shall include... the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;"
In particular I think the 10% rule may be overly conservative as most commercial music sites provide 30 second clips regardless of the length of the original. Perhaps we should amend this to 20% instead (which would allow 30 second samples for anything longer than 2 1/2 minutes). Kaldari 11:05, 19 June 2006 (UTC)

Keep in mind that we want all our content to be reusable by commercial entities as well as non-commercial. Deco 13:48, 19 June 2006 (UTC)

Kaldari, the 10% rule is not flexible. Please read http://www.utsystem.edu/OGC/IntellectualProperty/ccmcguid.htm. ~MDD4696 14:59, 19 June 2006 (UTC)

Those guidelines are for re-using music samples as parts of new multimedia works which seems significantly different from our use here, which is essentially to facilitate commentary and analysis of the original work (arguably one of the most uncontested realms of fair use). Regardless, I will go along with whatever most people think is appropriate. Defending such arbitrary rules by saying they are simply "not flexible" doesn't seem to be an especially compelling argument, however. Kaldari 22:39, 19 June 2006 (UTC)

BTW, if you go on to read the rest of that site, it says that the guidelines in question failed to achieve consensus support since many users "thought the Guidelines were overrestrictive". Kaldari 22:43, 19 June 2006 (UTC)

I think to call the rule "not flexible" is silly. It's not a law or any kind of widely accepted standard, and I can easily imagine cases in which it's very useful to sample a longer than 10% piece as a special exception. I think it's a good general guideline though, if you're not sure how long to make your sample. Deco 23:14, 19 June 2006 (UTC)

The guidelines aren't law, so they are flexible. However, the claim that there is no consensus isn't completely accurate... See this relevant point (emphasis mine):

"While only the courts can authoritatively determine whether a particular use is fair use, these guidelines represent the participants' consensus of conditions under which fair use should generally apply and examples of when permission is required. Uses that exceed these guidelines may nor may not be fair use. The participants also agree that the more one exceeds these guidelines, the greater the risk that fair use does not apply."

Also, Wikipedia and its editors do seem to meet the general description of 'educators' for the purpose of the guidelines. Wikipedia is a 'multimedia presentation' by their broad definition, so these guidelines should be taken seriously for our purposes. Could we violate them? Sure! But we might be playing with fire if we acknowledge the guidelines, and then choose to violate them. Thats a little tidbit a court might find interesting. Phidauex 00:17, 20 June 2006 (UTC)

It seems those guidelines were merely proposed and never actually agreed upon: "In late April, 1997, Bruce Lehman, Commissioner of Patents and Trademarks, publicly stated that the Proposed Guidelines negotiated by CONFU participants had failed to achieve consensus support. In May, 1997, at its third "final" meeting in Washington, D.C., CONFU participants concurred. None of the Proposed Guidelines would survive the comment and endorsement process that ended in May." Kaldari 00:30, 20 June 2006 (UTC)

[edit] Good Examples

In response to the concern that there were no 'good examples', I've created one as a good example. The article All Along the Watchtower about Bob Dylan's famous song now includes a sample of Jimi Hendrix's highly influential cover, released in 1968. The sample has been placed as an inline sample in the relevant paragraph, but also in a "Music Samples" section at the bottom. I wasn't sure which would be most appropriate, so I used both.

The sample is 23 seconds long (the song is 240 seconds), and was encoded as a stereo, 44,100hz Ogg Vorbis file, Quality = 0 (for an approximate bitrate of 64kbs). It sounds decent, but is clearly of a 'lower quality' than the original. It only uses 189kb of disk space.

Creating the sample made me realize how difficult it is to find a 24 second long section that includes all the major components of the song, the sing-song vocals, the slide electric guitar, and the famous solo. Some advice on picking the best sample might be helpful in the guideline. The first 30 seconds may not be the best 30 seconds!

Hope this helps as a sample. Phidauex 18:30, 18 June 2006 (UTC)

Great example, the sample seems low quality enough, but like this guideline recommends, It's better to insert the samples next to paragraph mentioning them to justify their fair use, instead of grouping them in the end of the article. CG 13:14, 19 June 2006 (UTC)

That's a great sample, and so tiny in size too - only a 30 second download for a modem user. I agree that some helpful guidelines on selecting the best sample would be great, although if a fan of a song is taking a sample, I think they'll appreciate the importance of this. Deco 13:57, 19 June 2006 (UTC)

One other note: single audio samples should use the listen template rather than the multi-listen template. Kaldari 14:36, 19 June 2006 (UTC)

Thanks for the tip on the Listen template. There are so many templates around here I get all mixed up on which one to use! Here is the same sample, formatted with the listen template:

As to the guideline suggesting putting them with the paragraph, I'm a little unclear about the rationales... In this case (All Along the Watchtower), the purpose of samples would be to allow a reader to compare the great differences in how the various artists have interpreted such a simple folk song. To that end, it seems like it would be helpful to have them all listed at the end, as well as inline, which is how I've done it. That way, when someone hears the the Jimi sample, and wants to compare it to the U2 sample, they don't have to go hunting for the next sample, they can just check the bottom. This would be analogous to having both inline citations, as well as more complete bibliographic information at the bottom, listed in both places, to support both uses of the information.

However, if fair use law is strict about this sort of thing, then I suppose I could remove the 'grouped' section at the bottom. Phidauex 17:24, 19 June 2006 (UTC)

In retrospect, I think I like the Multi Listen template better than the Listen template. The MultiListen template box floats and text wraps nicely, but the Listen template functions as a paragraph element, making it seem much less 'integrated' with the article... Phidauex 17:31, 19 June 2006 (UTC)

Oh geez, just ignore me. The formatting difference isn't due to the multi-listen template, its due to the {{Sound sample box align right|Audio samples:}} {{sample box end}} wrapper. Phidauex 17:45, 19 June 2006 (UTC)

All Along the Watchtower now has a complete set of music samples for the original version, as well as the three cover versions important enough to have their own section. Each sample is Ogg Q=0, and 10% or 30s (whichever is less - the DMB sample is the only one that is actually 30s). Each sample is listed only once, and is located next to the paragraph that discusses it. I'm using the sample box template, and the listen template, and have changed the template text from Audio Sample: to Music Sample:. This seems to be the most appropriate format, balancing integration with the article with easy identification of relevant samples.

I'm trying to make this article a good model for this proposed guideline, so any more suggestions would be appreciated! Also, I support the recent addition to the guideline of one sample per song recording as opposed to just one sample per song. That adjustment is critical for articles like this one and Amazing Grace where the educational componenet lies in the comparison of various interpretations of the work. Phidauex 21:28, 19 June 2006 (UTC)

I'm a bit unsure about per-recording versus per-song. On one hand, there are clear cases where the recordings are drastically different performances and really should all be presented. On the other hand, such a rule could potentially be abused to piece together a song from a number of nearly-identical but separately-recorded performances. Also, it should be clear that when comparing similar versions of a piece that samples should focus on the differences. But this might be best all left to editor discretion and common sense. Deco 21:34, 19 June 2006 (UTC)

While a song assembled from a composite of different recordings would be an interesting art project, I don't think its going to be an actual issue. The 'lower quality' requirement, for one thing, means it would not be very appealing to spend a lot of time 'reassembling' songs from various recordings, not to mention the bizarre quality to the final product.

However, I'd be willing to bet that nearly every song notable enough to demand its own page and samples will have been interpreted by a number of musicians, over a number of years (See Amazing Grace and All Along the Watchtower). Making it clear that it is OK to create samples from each recording for the purposes of comparison will make it a more friendly guideline. As usual, editor 'common sense' and discretion will come into play, but this is no different from all the other cases left to individual discretion. Phidauex 00:06, 20 June 2006 (UTC)

[edit] Straw poll

There's some concern that this proposal might have stalled, but there seems to be good consensus, so before we move forward with the process let's have a straw poll: policy, guideline, or oppose.

[edit] Template example in Guideline

I don't think that the instructions for how to use the multi-listen template are appropriate here. If that template is changed or deleted, no one will know to change the instructions here as well. Wouldn't it be better to just provide lisks to the appropriate templates? Plus, multi-listen is not a good example anyway as Template:Audio and Template:Listen are usually more appropriate. If we're going to give an example for multi-listen, we should provide examples of how to use the other 2 as well, including what they look like when displayed. Kaldari 17:48, 25 June 2006 (UTC)

I agree that we should provide examples for all the templates. To avoid duplication, we should either document their use only here and link here from the templates, or create templates to place in both locations. Deco 19:45, 25 June 2006 (UTC)

I don't think that Template:Audio is appropriate. It's too discrete, and it would be better if the presence of samples is clear. As for Template:Listen, it's good except that it can't be added next to the text, a principle which we agreed on to enforce fair use. CG 20:47, 26 June 2006 (UTC)

My thoughts... Template:Audio is too small.. Its good for pronunciations and things, but not so good for music samples. Template:Listen and Template:Multi-listen are what we should be using, as well as a sample box to 'float' the sample, and allow it to be positioned with the appropriate paragraph. Here is the code:

{{Sound sample box align left|Music sample:}} {{Listen |filename=Bob_Dylan_-_All_Along_The_Watchtower.ogg |title=Bob Dylan "All Along the Watchtower" (1967) |description=Sample from Bob Dylan's "All Along the Watchtower". From the album '''[[John Wesley Harding]]'''. |format=[[Ogg]]}} {{sample box end}}

See All Along The Watchtower for usage examples. I suspect Listen will get used the most, because it is better to have the samples spread out in the page, next to the relevant paragraph, than 'bunched' in one location. However, some cases might need a multilisten next to a paragraph to allow several samples that relate to that paragraph to be listed together. Phidauex 00:10, 27 June 2006 (UTC)

You're right, if we use the {{Sound sample box align right}} template it will work. Could someone add it in the guideline? Or I'll add it by myself later. CG 07:50, 27 June 2006 (UTC)

Added. I've also attempted to clarify the suggestion for template choice, and changed the example code to a music sample that meets the other parts of the guideline as well. Phidauex 17:44, 27 June 2006 (UTC)

[edit] "Relevant samples"

The current guideline says "music samples serve as tools for a better understanding of the article, so insert only relevant samples". It doesn't make it clear when a sample is relevant and when is not. Seemingly it's already agreed that not every mentioned song needs a sample, not even every song with an article (I actually agree that there are a lot of articles about songs which don't really deserve them). But how about recommending adding at least one sample to every article about a musician or musical style? Jogers (talk) 12:48, 26 June 2006 (UTC)

I think the question of relevance is best decided via consensus on a case-by-case basis. Deco 13:26, 26 June 2006 (UTC)

Yes, it should be clear that there should be consensus about which song and which section of the song should the sample present. CG 20:49, 26 June 2006 (UTC)

Sure, but I was thinking especially about some articles about the bands I like such as Kult or Akurat. They are certainly notable and well known in their countries but I guess they are not many people right now who watch or even edit these pages. If I propose to choose a sample to use at these articles' talk pages it is most likely that nobody will respond. I would be happy to know that I can choose and upload some samples that I find appropriate. Jogers (talk) 13:52, 26 June 2006 (UTC)

I was also thinking about how to encourage editors to upload samples to the articles about musicians because IMO it would significantly increase their value. What to you think about a template similar to the {{Needsinfobox}} encouraging to upload a sample and linking to this guideline for instructions? Jogers (talk) 14:17, 26 June 2006 (UTC)

[edit] Special:Unused media

I don't know where to post this, but I think that a special page called Special:Unused media would be more than useful for this project. The guideline specifies clearly that all uploaded samples should be immediatly included in an article. Could someone make this proposal in the appropriate place? CG 20:51, 26 June 2006 (UTC)

[edit] CSD criteria

Since the guideline forbids uploading samples that won't be used in an article, what do you think of a CSD proposal which allows the speedy deletion of unused samples? CG 08:11, 27 June 2006 (UTC)

There is already a section called "Images and Media", although most of the current actual criteria are geared towards images. I'll go over there and see what the feeling is about a refactor to accomodate all media. --james _{// bornhj (talk)} 08:26, 28 June 2006 (UTC)

[edit] Classical Music Samples

First of all, I should apologize for having come to this discusssion rather late, but after having read through the guideline proposal, and this page, I find no mention of guidelines for any music other than modern songs with a duration of less than five minutes. Many works of "classical" music (I'm using the term classical in it's broadest sense) are longer than an hour, some forms such as Opera and Oratorio can exceed three hours.

In a work such as an opera, does the 30 sec. limit apply to the entire 3 hour work, or could there be several 30 second samples from different sections to (for example) illustrate Wagner's use of a Leitmotif? If samples of several different sections are allowed, does each one have to come from a different recording? What about a long, single movement work such as Richard Strauss' Eine Alpensinfonie (which is nearly an hour long); is it still a 30 sec. limit, or may there be a few samples which combined are less than 10 percent of the entire work?

I don't know the answers to these, or many other questions as to how this guideline would be applied in regard to forms of music which are larger, longer and more complex than modern pop tunes. I don't know these answers because the proposed guideline does not even begin to address these issues.

While I admit that the use of music samples in classical genre articles is not currently widespread on WP, this is an area with great potential for growth. Unless the guideline language is adjusted to accommodate longer musical genres, any potential growth will be stunted before it begins. I therefore hope that these issues will be discussed, and that the guideline will not be adopted until these questions have been addressed. If you've read this far, thanks for your time. --Wine Guy Talk 01:23, 29 June 2006 (UTC)

You raise some good points. I think the guideline could still be clarified a bit, because I agree that its an important issue.

First, I'd like to point out that this guideline is only for music that is under a restrictive copyright license, requiring us to 'beg' fair use. There is quite a bit of public domain and copy-left type classical recordings out there that one could use the entirety of as a sample. University orchestras, for example, frequently release music for educational use. Here is a list of 'free' sounds already on wikipedia, which includes mostly classical selections: Wikipedia:Sound/list.

I think, for the case of an opera, one could use a 30 second sample from each song. When you listen to a 3 hour opera on CD, its not a single track, its broken up at logical points. Each of these chunks could have its own sample. Likewise, one could probably justify a sample for each movement of a multi-movement piece.

Unfortunately, the fact that it's an educational hassle doesn't change the legal requirements... Classical music copyright holders are as vicious as any other. Remember recently when John Cage's estate sued an artist who included a silent track on his CD labeled 'Silence', because Cage had 'composed' a song of the same title, which was 4'33" of silence. While I respect the art of John Cage, I'm disappointed in the frivolty of his estate's lawyers.

I think we will still be held to the same '30s of total sample time from each song' requirement with classical selections as we are with pop selections, in the cases where a free license alternative is not available (which they are, for many recordings).

Another alternative is MIDI files, which, for compositions older than 75 years, are now public domain. They aren't very expressive, of course, but for structural details, they can still be useful tools.

Anyway, I hope this helps clarify things a bit. Do you have any recommendations on ways to make the guideline more clear with regards to classical music, understanding that the legal differences aren't that different? Phidauex 03:12, 29 June 2006 (UTC)

Thanks for the response, Phidauex. I also think that a 30 second sample per movement/section of a longer work is acceptable under fair use provisions. But, I do understand very well how tricky copyright considerations can be; I nearly got sued by Boosey & Hawkes over an "unlicensed public perfomance" when I was still in college. If anything, classical music copyright holders are more vicious. It should also be mentioned that although a composer may be long dead, the performing edition (sheet music) may be copyrighted, so the number of truely free alternatives may be considerably less than you might think.

I should also point out that I'm not trying to make this out to be a classical vs. pop issue. Really, it's more about the length of a work; classical is a good example, but there's also Broadway, some jazz styles, and of course Tommy. I'm very interested that these types of considerations be worked out before the guideline is made official. To that end, I'll give this some more thought (and research), and see if I can come up with a slighty amended version of the guideline. I'll let you know if I have any bright ideas. Thanks again. --Wine Guy Talk 07:59, 29 June 2006 (UTC)

One way of addressing this that is already covered by the guideline is that samples of different movements could be taken from different performances of the same work. Although it's somewhat more difficult to compare movements when the performers are changing, it does simultaneously illustrate differences in performance styles. For arrangements that really are public domain, for example played directly from centuries-old sheet music, it would be great to convince some high school and university orchestras to assist in freely licensing their works. This wouldn't be too difficult at all for one-instrument solo pieces. Deco 11:36, 29 June 2006 (UTC)

May I suggest amending point 5 of the guideline to read as follows:

• There may only be one sample per song recording on Wikipedia; multiple samples are not permitted under our fair use policy. If a new sample is uploaded, the old one must be deleted. In the case of a multi-section/movement work such as a Symphony or Opera, the use of one relevant sample per section/movement is acceptable.

I believe that this clarifies the guideline, and adheres to both the original intent of the guideline, and the spirit of the legal interpretation on which the guideline is based. --Wine Guy Talk 19:17, 1 July 2006 (UTC)

That sounds fair to me, but I don't know much of anything about possible legal issues. Gyre 23:59, 1 July 2006 (UTC)

Sounds good to me too. Obviously we shouldn't have 30 second samples of *every* movement if we're not talking about parts of them in the article. --james _{// bornhj (talk)} 03:45, 2 July 2006 (UTC)

That suggested clarification sounds good to me, and reflects what I know about fair use guidelines. As always, samples will only be allowed when it is relevant to the topic. I might not include a sample of every movement in a piece, just as I may not include a sample of every cover of a pop song, or every sample on an album. The other parts of the guideline still apply. I've added your text to the guideline, pending any dissent. Phidauex 16:32, 3 July 2006 (UTC)

Yes, I absolutely agree that any samples need to be relevant to the article. Thanks to all for your thoughts on the subject. --Wine Guy Talk 18:53, 3 July 2006 (UTC)

[edit] Proposed CSD criterion

I proposed a new CSD criterion related to this guideline here. Please comment there. CG 19:46, 1 July 2006 (UTC)

I've suggested on the CSD talk page that there is already an applicable CSD in place for flagrant violations. Here's my reasoning:

• CSD I/M 7 Invalid fair-use claim says "Any image or media with a clearly invalid fair-use tag...can be deleted at any time."
• The music sample fair use tag says "The sample is short in relation to the duration of the recorded track..."
• Short is defined in the guideline as 30 seconds or 10% of the length of the original song...

Therefore, a copyrighted, fair-use sample which is longer than 30 seconds is not short, thus invalidating the music sample fair use tag and making it a candidate for speedy under CSD I/M 7.

If this is agreeable to others, perhaps we can work this into the guideline page. --Wine Guy Talk 21:16, 12 July 2006 (UTC)

[edit] It's a Guideline

The poll's result is 3/15/0 (policy/guideline/oppose). The proposal is now a guideline. Thank you all for you active participation in the discussions. The guideline will always be open for new proposals and amendments. CG 14:41, 5 July 2006 (UTC)

[edit] Lack of clarity

According to the preamble, this guideline is to do with the "usage of such samples need[ing] to comply with copyright law and Wikipedia's guidelines." However, most of the guideline is just concerned with copyright law and the text of the guideline doesn't make this sufficiently clear. The second point reads as follows:

"Samples must be of reduced quality from the original. A Vorbis quality settings of 0 (roughly 64kbps) is usually sufficient. To do this using Audacity, select Preferences under the Edit tab, and move the "Ogg quality" slider under File Formats to 0 before exporting the file in .ogg format."

Being an old-time Wikipedian I know that this concerns copyrighted music samples and free-content music samples. If we managed to obtain a free-content Wagner opera we'd want the quality to be pretty high both in the full files and samples (to illustrate motifs and whatnot). These guidelines need to be brushed up a little, it needs to be clear to the user that a guideline demanding low quality concerns copyrighted samples only. --Oldak Quill 13:00, 10 July 2006 (UTC)

It's clear that the quality and length limitations only apply to copyrighted samples, not free content; but it's only clear if someone has read this entire talk page. Since we can't expect everyone to do that, I've added the following to the preamble:

The limitations on length and quality described here apply only to fair use samples; free content samples are not subject to these limitations.

I hope that clears things up. --Wine Guy Talk 21:03, 12 July 2006 (UTC)

[edit] Backwards samples

What is the legal use of backwards audio clips created from copyrighted music? User:Bearingbreaker92 and I are willing to create and upload some samples to List of backmasked messages. Would the normal 30 second / 10% rule apply?

Also, would we be allowed to take the clips directly (without permission) from a website, or would that be violating the site's copyright? Λυδαcιτγ 16:52, 13 July 2006 (UTC)

These are derivative works - generally speaking, they're also original research and probably shouldn't be present. Since the publishers retain the right to create derivative works, such audio would have to fall under fair use to use here, and I'm not sure exactly what conditions would apply, but the same rule as for forward clips is probably a good starting point. Deco 17:29, 13 July 2006 (UTC)

[edit] Exceding fair use in sample length or in number of samples for album

What do we do with samples that are over 30 seconds in length? Is there a simple answer? Example Image:Moloko - Sing It Back excerpt.ogg where the sample is 50 seconds in duration, and the uploader has taken a stab at who owns the copyright but isn't sure.

They would be considered as copyvios and should be speedy deleted after notification to the uploader or the talk page. CG 13:42, 16 July 2006 (UTC)

Thanks. The majority that I've noticed were uploaded by the same editor, who says on his user page that he has left Wikipedia, however I will notify any that I detect. Rossrs 13:54, 16 July 2006 (UTC)

Also is there a limit on how many songs from one album can be sampled? Example Hi Hi Puffy AmiYumi (album) links to a sample for each and every track, and I'm sure that they are not all notable and not all relevant. But how should it be culled? thanks Rossrs 12:03, 16 July 2006 (UTC)

[edit] Audio from television episodes

I've used the Music Sample guideline to upload audio from a television series, The Force: Behind the Line. I've uploaded two audio samples (1, 2) in order to represent the narrative style and use of background music of the program. In order to ensure fair use I've included detailed rationales on both of the audio samples, and there is related commentary in the article to justify their presence. Is what I've done okay? Also, is there a more suitable licensing template than {{Music sample}} that I could use? Tntnnbltn 05:48, 30 August 2006 (UTC)

[edit] Where does 30 seconds come from?

Where does this number come from? Is there something wrong with a minute sample of a 70-minute work (Beethoven 9)? For example, at Timpani, I have specifically chosen samples to contextualize the points I am trying to make. Cropping the samples to meet the arbitrary 30s requirement would diminish the educational value of the clips. The context the clip is being used in should be considered. – flamurai (t) 04:25, 27 November 2006 (UTC)

The 30 seconds limit only apply to fair use music samples. As far as I know Beethoven's works aren't copyrighted. Jogers (talk) 10:14, 27 November 2006 (UTC)

The music is probably not subject to copyright but the recording may well be, depending on which recording is used. Rossrs 10:26, 27 November 2006 (UTC)