Publishing music with MuseScore soundfonts/MIDI

• May 13, 2020 - 21:18

Is it legal to publish music using soundfonts or MIDI exported off of MuseScore? I've found that it provides a small, but free array of sounds that are easy to use, and am wondering if it's okay to use them in my songs. Thanks!


Comments

If the score is copyright free, the MIDI surely is too, it doesn't use any sounds, but gets interpreted and sound at the target machine only.
And as long as the soundfont used for WAV, MP3, OGG or FLAC export are copyright free (and the ones MuseScore provides, preinstalled or as extensions, are), then the sound export is copyright free too

A similar question just recently came up on musescore.com, see https://musescore.com/groups/improving-musescore-com/discuss/5065621

In reply to by Marc Sabatella

I’ve posted this on .com already, but here’s the tl;dr form:

A waveform (MP3, …) export is generated by a program (like a compiler, here MuseScore) taking a source code (the score) and a library (the soundfont, “software” in lowercase in the licence) and combining them into a work. The resulting work is not influenced by the licence of the compiler, but it is covered by both the licences of the score and the soundfont.

The soundfont is MIT licenced. This licence requires one to: subject to the following conditions:

The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.

The resulting waveform contains a substantial portion of the soundfont (which is a collection of samples plus some processing declarative instructions), so therefore the text must be included. Incidentally, “Software” (in uppercase) is defined as the work itself (“software” in lowercase) or the “associated documentation files”.

So either embedding the licence information as an ID3 tag or placing it along the file whenever it is redistributed is a must.

On my site I currently place it alongside the files. I originally intended to put it into ID3 tags, but then I found out that MuseScore does not retain all metaTag tags in e.g. MusicXML output, and the metaTags are where I put licence information of the score, so I just collect them all. (They are currently found in the *.meta.xml files; making this more user-friendly and displaying them in the browser is the next step.)

But the soundfont most certainly is under copyright, under multiple ones at that. Start reading at “Licence:” here https://people.debian.org/~tg/sf/musescore-general-soundfont.copyright (i.e. skip the first three paragraphs, they apply only to Debian packaging). This is the information to reproduce.

In reply to by mirabilos

Can you point to the landmark legal ruling that audio file generated using a soundfont constitutes "all copies or substantial portions of this software"? This would be earth-shattering news, it would mean every piece of music recorded with practically any synthesizer would be considered a derivative work. No judge in their right mind would make such a ruling, and indeed, I submit it has never happened and will never happen.

In reply to by Marc Sabatella

Not right now, busy with dayjob, but this is common understanding, and especially was assumed when debian-legal looked at MIDI synthesisers and soundfonts (TimGM6mb for example is GPLv2-only, so it could not be used with nonfree, GPLv3 or most Creative Commons MIDIs to create a legally redistributable MP3/WAV/OGG/…).

But if you look at it: if you make a score with a note, say A₄ in the harpsichord, and then render that, the resulting WAV is identical (given no other filters, etc.) to the A₄ sample in the soundfont (called “Harpsichord A3”), which is under the MIT licence from Fluid (R3). So this conclusion should be trivial.

In reply to by mirabilos

Except that now it is in a form that is no longer usable for its original purpose, that of being a soundfont. That is, the MP3 is not the soundfont, it is a different type of work. That's part of the common sense distinction I am making here. It seems plain to me that the intent of the agreement is to say that the given text must appear on actual soundfonts derived from it, not on audio recordings made from it. I suppose the debian folks can be as conservative as they like, but there is no reason to scare ordinary users into thinking someone is going to come after you for publishing your audio made with MuseScore and its default soundfont. You absolutely positively will not be sued, and if someone were stupid enough to try, it would most likely be tossed out instantly before it ever reached court.

In reply to by Marc Sabatella

If you create a document from cliparts, it’s also no longer the clipart library.

If you create an executable from object files and libraries, you cannot use it as the libraries any more either (e.g. MuseScore is not qtsingleapp, but it contains qtsingleapp).

If you create a collage from excerpts¹ of original works (photos, musical pieces, etc.) they are also no longer the original works.

In all these cases, however, the result is a derivative work of the original work(s), and therefore at the mercy of the licences on the original works.

① very small pieces excepted of course, let’s assume they are sufficiently long enough

I cannot even begin to comprehend why people would think otherwise. I now can understand why the collection societies, music industry, etc. are attacking so harshly. Perhaps someone should sue for the MIT licence terms (AIUI it already has been held up in court, so I know this is possible)…

… why am I even bothering? This disturbs me.

(I am bothering because I wish for MuseScore to stay legal, so I can still use and endorse it.)

In reply to by mirabilos

Collages containing recognizable pieces of copyrighted material that lack any sort of license agreement is one thing. An audio file built using samples from a soundfont with a clearly worded license is another.

I'm happy to see someone tweak the wording of the license agreement is it helps you feel better about it, but again, to me it is plainly obvious what the meaning is, and no one has to my knowledge ever challenged this or even seriously questioned it before.

Not sure what you mean about someone "attacking so harshly" - attacking whom, over what?

In reply to by mirabilos

Funnily enough, the soundfont developers and the MuseScore people understand this. This is why we’ve been collecting as many samples as possible under CC0, as to be able to eventually build a soundfont under CC0, which then will allow freely licencing the resulting waveform under any terms you desire.

In reply to by mirabilos

I too have studied copyright issues for decades. But sure, neither of us are lawyers, and we live in different countries anyhow. You can choose to worry about this if you want, and certainly I applaud the desire to improve the soundfont further both in terms of quality and in terms of using CC0 licenses.

But I will continue stand up any time and someone advises a user they do not have permission to create music with MuseScore. You don't know that; it's just an untested theory. Until a court ruling proves this is a problem, I'm going with the overwhelming evidence and common sense logic that says this is not true.

In reply to by Marc Sabatella

“advises a user they do not have permission to create music with MuseScore”

I don’t. Did you even read what I wrote?

I said that, when they want to redistribute the MP3, OGG, WAV, … exported from MuseScore, they have to honour the licences on the soundfont as well, and the one used by the default soundfont merely requires a certain text to accompany it. That’s it. Otherwise, full freedom.

In the same vain, I have to intercede when someone tells people there was no copyright on the soundfont, because there most certainly is.

In reply to by mirabilos

Sure, I read "music generated using it has a chance of being a derivative work", but you're right, you did also say elsewhere in a separate post that one could get out of it by including a copy of the agreement. So, I will amend my statement to say, I will stand up any time someone adviser a user they need to attach a license statement when distributing their audio.

Of course, you are right there is a copyright on the soundfont, just as there is on MuseScore. It's not a particularly important distinction to make in most practical contexts, but sure, it's obviously literally true.

In reply to by Jojo-Schmitz

It’s virtually no effort to add that statement. It could even be embedded in an ID3 tag.

Not doing so is a copyright violation, which, when done commercial, is even a felony under German copyright law, independent of whether the rights holder sues you over it or not.

I believe the musescore.com site also must include this metadata.

In reply to by Jojo-Schmitz

Thankfully I had researched this last month for another reason:

>>(b) violation is a statutory crime even without prosecution.
>
>§ 106 UrhG (use of a protected work without permission), which in
>itself is indeed only prosecuted on request, EXCEPT if § 108a also
>applies (doing so commercially). See
>http://www.gesetze-im-internet.de/urhg/BJNR012730965.html#BJNR012730965…
>and the following paragraphs (§ 108a is excepted from § 109).

WAV files are RIFF so one could embed it as an INFO chunk or something. I’m not sure about the Ogg container or however FLAC is provided outside of it, but the technological side is not that hard to solve once there is willingness. I started with MP3 because that’s most common.

In reply to by mirabilos

It's true it's easy to attach the statement, but the statement "not doing so is a copyright violation" remains only your opinion. And my statement that the actual intent seems clear is my opinion, I'll grant that. So until a judge rules on this, I guess we'll just have to agree to disagree, and we can just keep standing up as we see fit.

In reply to by bobjp

I wouldn't discard mirabilos' remarks so swiftly; whilst not a lawyer, he is quite the expert on licensing issues.

In my reading, I'd say he is technically correct and if it came to court, you'd have a decent chance of being placed in the wrong if the judge chooses the strict interpretation.

On the other hand though I'm also extremely confident that nobody at MuseScore will pursue and enforce the license of MuseScore_General in such a manner. So yes, I also tell people that MuseScore and their included soundfont comes with no strings attached regarding the composition your create and render with them.

In reply to by jeetee

I agree that what he says isn't completely without merit. But to me it is strongly biased by a very particular "purist" perspective that I don't believe holds much real-world relevance. That is to say, actual case law tends to be rather more practical than that, at least in the US. The intent of the license in this context seems clear to me as being meant to apply to derivative soundfonts and not to musical audio files, as indeed that's the established norm in the audio sampling industry. And the lack of even the slightest attempt by the relevant copyright holders over the past few decades to assert otherwise would likely be taken into consideration. Again, I am confident no judge would rule in favor of someone claiming at this point that musical audio files should be considered "copies" in this sense.

But that's just my opinion indeed.

In reply to by Marc Sabatella

It seems to me that the license states that you can't take the item as is and redistribute it (or modify it) and call it your own without adding the terms of the license to your product. It also specifically mentions having to do that only if you are charging for the end product. That's the way I read his link. I would think that if it applied to any use of the product, then it would have said that up front, and not at the end. It seems to me to be only logical that calling MuseScore using a font as redistributing the original software, is a big stretch. If that were true, no one could use any software to produce anything without including a virtual ream of licenses.
It seems to me that MuseScore does not modify and reissue the font. It does use it within the boundaries for which it was intended. MuseScore does not change the font any more than I do when I copy the font to a different computer.
I don't think it takes a lawyer or court to understand that.
Yup. My opinion.

In reply to by bobjp

“It also specifically mentions having to do that only if you are charging for the end product.”

It doesn’t.

The MIT licence is well-understood in the FOSS world.

“when I copy the font to a different computer”

You have to include a copy of the licence then as well. Otherwise, in the absence of a licence, copyright law forbids you from copying¹ the font.

① except for private backup copies, etc. of course, but you cannot distribute these either

In reply to by mirabilos

What is the difference between downloading a font (which contains the licensing information),and copying the font to another location (which also includes the licensing information).

This is from the license for Virtual Playing Orchestra:

"If you're using this free orchestral sample library to make music, then licensing is easy. You can do whatever you want with the music you make, even sell it commercially.
If you want to modify, repackage or redistribute the library, licensing is a little complicated since different licenses apply to different parts of the library (see the source sfz files and see below) but generally, you can safely copy, redistribute, modify whatever you want provided appropriate credit is given to creators of the content used or at the very least, link back to my site which gives credit to the creators of the various content of this sample library and that any derived samples or library is kept for personal use only or is given away for free."

This says to me that I can freely use the font to make music without including license information. I do have to include it with copies or modifications. Since the license information is included with the font, I can therefore copy it to another computer.
You seem to make no distinction between use of the font to make music and redistribution of the font. The above quote (and others I have read) clearly does.

In reply to by bobjp

“This is from the license for Virtual Playing Orchestra”

You said it yourself: that what you cited applies to VPO.

The MS_General soundfont has a different licence, which it MIT. The MIT licence is a software licence, and therefore does not make a difference between using the library (soundfont) to create a program (music) and redistribution of the library (soundfont).

I never said the MIT licence was well suited for fonts, sound or typographic. But embedding typographic fonts into PDFs at least retains the licence information from the fonts (just tested a 3.2.3 exported PDF, it does).

I had already started working on making this automatic for soundfonts as well, and copy the metaTags from the score, in MuseScore itself. (It’s quite hard to figure out which soundfonts are actually used if multiples are loaded, but I could just do that for all loaded soundfonts.) But I’m reluctant of pumping man-weeks of work into this if people are not only not interested in it but actually against licence conformance.

In reply to by mirabilos

For the record, I have nothing against dotting i's and crossing t's in this particular way. I can see why from the perspective of wanting to produce a clean distribution with respect to licenses, it makes sense. And I can see why it's preferable to include soundfonts licensed in better ways.

In reply to by mirabilos

OK then, this from the MS general font license.
"including without limitation the rights to use,
copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the
Software is furnished to do so, subject to the following
conditions:

The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software."

This says to me that we are allowed to use the software. But it says nothing about the result of using the software. I would think that that would be spelled out.

In reply to by bobjp

You'd think that, yes. But as has been pointed out a rendered score using that soundfont would under strict interpretation (which is common in the FOSS world) be considered a "modified copy" of the soundfont; thus should include that notice in order to be compliant with the license.

In reply to by jeetee

I believe in copyright. But, sorry, this is a silly interpretation of the license. If this is so, the license needs to be reworded. Common interpretation doesn't mean it's correct. The license says nothing about what happens to the result of using the software. There is a long list of situations, but not that. I find it difficult to follow the necessity for including license info in an mp3 made with free software and a free font. Especially based on common interpretation. What is the point of any software? Is it not to use it to obtain some kind of result? It seems to me that in a license that specifically lists things that can be done to the software, that it might at least say something about the result of actually using the software. Whatever kind of software it is. I'm not sure how many different ways I can ask this question. Does the sound file contain any of the font code? Does the sound file modify the font code? Can users of the sound file modify the font code? Does the sound file redistribute the font code itself? I'm having a hard time logically answering "yes" to any of these questions.

I have no problem with the inclusion a notice in within software. But I find no justification for including it in any result of actually using the software. This, as I read the license, is at best a separate issue. Perhaps I'm just stupid. I can't see how anyone would come to the "common interpretation" based on the wording of the license.
I submit the above with all due respect. But it's just bazaar enough for me to not be interested in using MuseScore any more. Not that that should mean anything to anyone.

In reply to by bobjp

The soundfont is a piece of audio, which is combined with the score into another piece of audio, so OF COURSE it’s a derivative work.

But you don’t seem to understand that the licence for the current soundfont cannot be changed. It’s got multiple contributors and is a derivative work of another soundfont (Fluid (R3) Mono) which was released by just another person (taking work from two more people), and which is a derivative of just ANOTHER soundfont (Fluid (R3) GM), which is comprised of the work of fourteen (14) people.

So there’s at least eighteen people involved, some of them only known by their first name… good chance with that.

There’s a reason “we” are looking for CC0-licenced samples and asking for new contributions to be published under that licence. That will allow those with skill to create a new soundfont (which Daniel alluded to was going to happen for MuseScore 4; I wonder what name they’ll use for that, maybe MuseScore Commons?) under CC0, which then allows you to licence the resulting waveform any way you want.

Be thankful the current soundfont is MIT licenced, not CC-BY-anything… now those are attribution nightmares.

In reply to by mirabilos

So is the font software, or is it audio? I understand that a resulting audio file is a set of instructions for an audio player to render sound. But is there anything in a resulting (mp3 for example) audio file that can be traced back to the original font? If so, that's one thing. I'm not interested in not honoring the license. I'm just trying to figure out how an audio file can be considered software if the file is no longer identifiable with the font. Sure, manipulate the original font and reissue it with the license. That's a derivative work. It's still a font. Simple enough. The new font can be used to create an audio file. It seems to me that if the audio file no longer contains anything to identify the original font, then that is not a directive work, but something else.
No doubt the mp3 is just "understood" to be a derivative work. Which I fail to understand.

In reply to by bobjp

> But is there anything in a resulting (mp3 for example) audio file that can be traced back to the original font?

Yes. The word “software” in the licence text is… unhelpful here (but there is legal precedent, even in America, for software licences to be applicable to non-software as well). Please substitute mentally the word “work”.

In this specific case, there are “samples”; for example, the Harpsichord instrument has a sample for A₄ which is an audio recording of how pressing that key sounds. This ends up to 100% (unless very short, a 64th note or so) in the resulting audio file. And this audio recording is what 14 or so of these people claim copyright on. (Frank Wen, ChurchOrganist and s.chriscollins can additionally claim copyright on the processing instructions… declarative data that, for example, says which part of the original sample to “loop” when the key is pressed for a long note, especially in e.g. wind instruments, and which pitch correction, panning, etc. to apply.)

So, yes, the MP3 contains audio which is a direct derivative of the audio that the soundfont is comprised of.

In reply to by mirabilos

"So, yes, the MP3 contains audio which is a direct derivative of the audio that the soundfont is comprised of."

That was not my question.
If the owner of the harpsichord font hears some random mp3 without knowing how it was made or if his font was used, is there information in the mp3 that labels the harpsichord sound as his? I mean, he could say that it sounds like his. it would be harder to tell if EQ and reverb were applied. How would he go about proving the font used was his?
And why on earth would he care as far as the legitimate result of using free software goes. Otherwise, charge a small fee and write a better license. Except that I can see that as being even more hassle for small operation.

In reply to by bobjp

“it would be harder to tell if EQ and reverb were applied” we’re not going to try and undermine copyright here.

“How would he go about proving the font used was his” I’d assume there are methods, for example synthesising from his soundfont with the score and comparing the result comes pretty close. I only know such methods exist for software, but have no experience with digital signal processing.

“And why on earth would he care as far as the legitimate result of using free software goes” EXCUSE ME‽

The acknowledgement of being mentioned is why we publish Open Source and Free Software. If you wish to violate that, you can pretty much fuck off. If you can’t even honour that one little request then you do not even deserve to use MuseScore or the soundfonts.

In reply to by mirabilos

I am not interested in subverting copyright. I understand it and have worked with it more than you know. I am merely engaged in an interesting discussion about the need for including the license with an mp3. I am told that it is “understood” that the mp3 is a copy of the licensed software (font, work). I ask how anyone knows what font it is a copy of. No answer. But I am told that the designer of the font (software, work) wants recognition for their work. I understand that. But what about the 14 other people that I am told may have worked on the font? Where are their names? Are they lessor contributors?

I ask how someone could use my mp3 to create a new font. I get a plausible answer. That answer might work for a single line, single instrument mp3. I am not sure it would be possible for a full orchestra mp3.

Let us say, as is “understood”, that the mp3 is a copy of the font (software, work), and therefore software (font, work) on its own. I did the work of manipulating the font (software, work) into the final form of the mp3 (font, software, work). Therefore, I am the author of the of the software (mp3, font, work). As the author, I should get to put my name in the license. It seems I might not have to include the names of other people because they are lessor contributing authors most of which are not listed to begin with. I want recognition for my work. Otherwise, it is “understood” I can colorfully invite you to not use it.

This is all hypothetical, of course, because of all the mysterious things that are “understood”. I do not mind if I am proven wrong. I get that whether I understand “understood” is not important to anyone but me. I am just asking some questions. For which I am formally invited to leave. Interesting.

In reply to by bobjp

> I ask how anyone knows what font it is a copy of.

Ideally by the metadata of the MP3, or the accompanying documentation. And to add such is the licence condition in question.

> But what about the 14 other people that I am told may have worked on the font? Where are their names?

In the same copyright notice. Basically, anything below “Licence:” in https://people.debian.org/~tg/sf/musescore-general-soundfont.copyright is the copyright notice to reproduce.

> Are they lessor contributors?

What is a “lessor”?

> I am not sure it would be possible for a full orchestra mp3.

The ability to do this for one case at all is sufficient to point out how it’s a derivative work of the samples in the soundfont.

In reply to by Jojo-Schmitz

Yes, correction needed. Lesser. less important. My point is that I seem to remember mirabilos mentioning that some of the fonts used by some people listed in the license might be based on the work of others not in the list. My question is two fold: Why are they not in the list, and can I add my name to the list? The mp3 is a direct result of my work. Much like the font is the direct result of their work.
I'm just asking as a rhetorical question. I find it interesting. No need to get defensive.

In reply to by bobjp

No, they aren’t (you misunderstood me, the link I posted contains a complete-as-we-know-it-and-as-the-authors-provided-to-us list), and yes, you can, but only for your resulting file of course.

This is how the list happened anyway: someone made Fluid; someone took it, added significant work and added themselves to the list and called it FluidR3_Mono; someone took it, extended it and added to the list… you get the idea. This is how FOSS works.

In reply to by mirabilos

I would love to read the argument that the raw samples inside a soundfont constitute a "work," and that the further arrangement of these sounds is thus a "derivative work." The argument would struggle to bear a relation to the patentable arrangement and orchestration of a musical work. The problem is that the original samplers have slowly ceded their rights through time and lack of enforcement, partly due to the difficulty of a method of proof. ( I once defended copyright on a college panel during latter Napster years and I am more than happy to credit the creators.)

In reply to by ramblinj

have slowly ceded their rights through time and lack of enforcement
I gather that this seems to be a mostly American thing that you should actively enforce your rights; that isn't true in at least one other country (Belgium :) ) so I imagine that to be not true in other regions as well.

In reply to by mirabilos

There was a time under US law where works could lose copyright status if not actively defended, but that went out even before we joined the Berne Convention I believe. But that' a separate matter from establishing the intent of a legal agreement; here actual behavior can and does play a role.

In reply to by ramblinj

What exactly is the work? I sample minor thirds across two octaves and then list from lowest to highest pitch the wav files under an instrument name and then call the preset "fantasia." Would the "work" be the sequence of minor thirds such that the derivative work would be any rearrangement of the original sequence?

No, the work must have to do with the raw quality of the sound itself, its use and further processing, The term "derivative" here being a recognition of a chain of custody. The unions have thought about this and not just in terms of recruitment. If something is commercially successful enough they will collect monies from major media buyers regardless of the impossibility of tracing a chain of custody. They do this in recognition of the fact that there are humans behind these sounds.

In reply to by ramblinj

I have spent much of the day reading legal websites on copyright and licensing. In particular, MIT.
Including what constitutes a derivative work of computer software. AIUI, software is a set of instructions that when acted upon by a computer, yield a certain result. If that program is altered, expanded, improved, etc., the resulting software is a derivative work.
Nowhere is there any mention of the result of using the software. Results like: A Power Point presentation, Word doc, or an mp3.
Not only that, but as I think of it, I am having difficulty considering the MS General font as being software. Other fonts I have looked at contain some instructions for the computer on some aspects of the sound files. But as near as I can tell MS General is only a collection of sound samples. Data. Data that MuseScore knows how to interact with. It seems a stretch to call a data pool, software. It seems an even further stretch to identify an mp3 as software any more than a Word doc could be so identified.
One site suggested that the derivative work angle is not a good fit for works of art, music, and even movies based on books.
And the MIT license is, of course, made for open source software. But is very thin as far as derivative works go. And probably not a good fit for music software. If it is software.
The problem is that there is little universal agreement on some of the fringe areas.
Marc has already gone into who owns what in a derivative work in other threads.

In reply to by bobjp

  1. you’re still too hung up on the word “software”; please mentally substitute the word “work” when thinking about the MIT licence in soundfont context (this is what courts do when looking at software licences in non-software contexts)
  2. the MP3 is not the result of using the soundfont-software. It is the result of using the synthesiser-software to combine the MIDI-data and the soundfont-data.
  3. we’ve known for years that the MIT licence is not optimal for soundfonts, BUT IT’S ALL WE HAVE since it’s the one we received the current soundfont under, so unless you use a different one, just fucking honour it, thanks

In reply to by bobjp

I am having difficulty considering the MS General font as being software.
Unfortunately, with the current license, that technically is exactly how it must be viewed.

Everyone in this thread agrees that it is not a good choice for SoundFonts. But it is at this point in time still "the best we have available. And because we believe it to not be a good choice currently there is the ongoing effort of constructing a decent enough (quality wise) CC0 soundfont for future inclusion.

This thread was never about what it should be; but how to (ideally correctly) deal with the current facts.

In reply to by jeetee

Guys, please. I know it looks like I'm being unreasonable. I am trying to deal with the current facts. I believe that the wording of the MIT license is fine regarding fonts, as such. Sure, if I take the font and add to it, improve it some way, copy it verbatim, and so forth, then I need to include license info. As I read the license, I can see no way that it applies to the result of using the font. So far, I can find no legal precedent for doing so. So far, the same goes for other similarly worded licenses. Even if the mp3 is considered a derivative work, it is best, and in fact already is, covered by a different license.
I would expect that by now someone would have posted some legal precedent proving me wrong. I am happy to be proven wrong. This is how I learn. Is it an opinion that I need to include the license with the mp3, or is it a verifiable by legal decision, fact?

In reply to by bobjp

“result of using the font”

And this is where you need to wrap your brain around the wording.

The resulting audio is not the result of using the soundfont, but it’s the result of using the synthesiser, which combines the soundfont with the score into audio.

Therefore, the resulting audio is not “output of the soundfont” but “modified copy of the soundfont”.

In reply to by bobjp

I'm not aware of any such statements made by those; but even then, the only conclusion you can link is that they chose the license inappropriately, it doesn't nullify them.
Now if you could get approval of all of them, then the font can be relicensed.

Again, this is not about what it should've been, it is about what it is.

As I have stated before, one point where in this instance my stance differs from that from mirabilos is that I also believe that if it ever came to a court case, you'd likely can argue your case decent enough and the ruling will likely favor the allowance in that specific case. (Again, my opinion and guess).
But that doesn't make the fact of the license any less true; and that to abide by the license you should indeed include it in the derived product.

In reply to by jeetee

I would say this describes my own stance pretty well too. I think all of the following are true:

1) a literal reading of the license, absent any knowledge of prior precedent with regard to music licensing specifically, could be seen as requiring the statement to be included in any audio files generated from it
2) because of this, it would be great if MuseScore automatically included it in the metadata of the audio files it creates
3) also because of this, it would be great if we could find/develop a new soundfont, or somehow get the necessary agreements from all parties to relicense the current one

However, I also believe the following:

4) it seems very unlikely that the intent of the developers of the soundfont was to impose this restriction
5) it seems very unlikely they would choose to try to enforce that interpretation at this point
6) it seems very unlikely a court would agree with that interpretation given the conventions and unique attributes of music licensing

So for me the bottom line is still, individual users have nothing to worry about. If they wish to include the license statement, they are welcome to do so, but are only obligated to starting the day a court actually rules this way. Until then, it's just a recommendation based on an untested theory of how a court might rule - a theory based more on experience with other software rulings than other music rulings, it seems.

In reply to by Marc Sabatella

One last thought. I hope.
Believe me, I really do get it. If I make a copy of the font, I need to include the license. If I modify (derivative work) the font, I need to include the license. No problem.
Everything else seems like a problem to me. The devil is in the details, it would seem.

“The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.”

This says to me that there are two situations that require inclusion of the license.
1. A copy of the software. That would seem to me to be the entire font, because of the second half of the statement.
2. Substantial portions. Which I guess means what? ½, 3/4, 1/3?
I don’t think my mp3 using the harpsichord samples qualifies in any way for license inclusion because it is certainly not a copy of the full or substantial part of the font.
It is difficult for me to believe that the developers of the font ever intended for the license to be included with an audio rendition. They aren’t stupid. What other possible use could there be for a font but to make an audio file? How could they have dreamed that it would be necessary to add a statement about not needing to include the license in an audio rendition of the font? Which they could have done.
Include it in the meta data? I don’t think it’s MuseScore’s job to do that. They are not the end user. Besides there are plenty of fonts that don’t have the same requirement. As the end user, I would have to do that. Even if I have a clue about how to do that, the person listening to the file would have no idea it was there. That is not a reason to not include it, I know. So, I include a text of the license. There, I’ve satisfied the license. But, how and why would any court waste time on this?
Besides, MuseScore has to load the font into the synthesizer. Is the license included with that copy?

In reply to by bobjp

> Substantial portions

Anything that falls under copyright protection, i.e. nothing trivial.

> that the developers of the font ever intended

Nevertheless, they chose an established FOSS licence. Ask them, if you want to know why. In the meantime, I’ll hold up FOSS community standards.

> Besides, MuseScore has to load the font into the synthesizer

These are transient copies required as part of transmission or usage and generally exempted.

In reply to by mirabilos

Music licensing is, as I said, unique. The concept of digital vs analog copies, the notions of pubic performance and of broadcast, and the specific precedent set with respect to how other synthesizer and sound libraries are licensed, the way "fair use" has been interpreted over the years - all of these are ways music licensing is different from software licensing. Not more or less strict overall, just different.

The clearest summary I can perhaps give has to do with the traditional way "fair use" is evaluated under US law. There are four well-established criteria used to evaluate whether a given usage is considered permissible. They have to do with the purpose of the new work, nature of the original work, the substantiality of the material taken, and - the most important according to most analyses - the effect on the market for the original work.

The first and last of these alone are more than enough to yield a very clear and obvious result. Compared to the original soundfont, the MP3 of one's original music would unquestionably be seen as "transformative" to a very great degree - far greater than any case I am aware of that was ever ruled the other way, and indeed less so than many that have previously been judged "fair use". The original was a collection of individual pitches with no musical value whatsoever; the audio file is some piece of music - assuming it is not just a rendition of the individual pitches played sequentially for easy sampling. Also, the effect on the market would trivially be shown to be zero. And I mean, very literally, zero effect on market whatsoever. That slone would probably result in this never coming to court.

So even before one comes to the question of the nature of the original work or the substantiality of the material used, it's pretty much a done deal. But still, let's say that despite the overwhelming reason to believe such a suit would never be brought up and would never reach court if it did, somehow a judge finds himself hearing/reading arguments. For a soundfont that consisted of unique and easily identifiable samples - fragments of speech, musical snippets, creative synthesized waveforms that are unlike any other - then indeed, use of a recognizable portion could be problematic. But that's not the case here - these samples are incredibly generic-sounding. And that is indeed true due to the very nature of this soundfont - it's intended to be generic in this way, as are virtually all GM-compatible soundfonts. The nature of the soundfont is to provide generic-sounding instruments whose sole purpose is to be transformed into music by the creative work of the person creating the audio file.

So, that's just the fair use analysis under US law, but again, the idea of what a soundfont is and how it intended to be used is just commonplace and well-known that you couple replace those specifics with those of pretty much any other jurisdiction and come to a similar place. When it comes to sampling music, yes, anything recognizable is indeed potentially problematic. When it comes to synthesizers, sample libraries, and soundfonts, this naive comparison won't hold water.

In reply to by mirabilos

Like I said, that's just an example, but I do believe similar consideration would apply in other jurisdictions, because the concept of a synthesizer and a sample library and how these are traditionally licensed is universal.
I just gave what was a very specific example of how one passage in US law would apply, because it was easy to see what the principles are here and because I am pretty intimately familiar with them and thus didn't need to read entire treatises on Belgian law or whatever to speak to.

Still, I think it would likely more come down to what I suggested initially: whether the originator of the soundfont suddenly came out of a decade-plus coma and decided he really wants to claim his intent really was to insist on something that has never once in the entire history of digital music ever been claimed before (that people should include a license statement in the music they create using his own General MIDI soundfont), and whether a judge would accept that and be willing to establish this new landmark precedent that would shake the world of recorded music to its core.

In reply to by bobjp

This is normally the intent of Open Source people choosing such a licence: to get attribution when their stuff is used.

I talked to someone, Ethan Wiener I think, for reusing some stuff under CC0 in MS_General, and while he agreed to CC0, he would still prefer for attribution to show up in the MP3 metadata, so it’s also searchable and discoverable on sites that index MP3s, for example.

So, yes, this one condition the MIT licence has is the full intent of at least one of the contributors to the soundfont.

In reply to by mirabilos

Interestingly enough, there is an Ethan Winer (your Wiener ?) listed in the MS General license who contributed Temple Blocks. On his website he lists Temple Blocks (and other fonts) as free to use. "These files are made available for you to use in any way you'd like, royalty free, including for commercial projects, with no attribution required. Enjoy!" So I'm a bit confused as to his actual intent. If it is the same person, you say he told you one thing but his website says another.
Did each person who contributed agree to the MIT deal, or did one of them just decide to add it?

In reply to by mirabilos

I would also observe there is a difference between "Since you asked, I would prefer MuseScore add the attribution automatically to the MP3 metadata since it is easy and convenient to do so" and "From the beginning, I have always fully intended to sue anyone who releases an audio CD or cassette tape that does not include this statement printed in the liner notes".

As I keep saying, I fully support efforts to improve MuseScore's compliance with the letter of the agreement and to find alternatives that are not so encumbered. But I continue to argue this has little bearing on the question of the legal burden for individual MuseScore users (it demonstrably does not apply in the US), and that it is misleading and unproductive to tell users they need to do take any specific action with respect to the audio files they create.

In reply to by bobjp

But I find no justification for including it in any result of actually using the software.

So the following use case scenario should be entirely valid according to you?
Create a score containing an instrument and notate out each note of its range with enough rest behind it to allow the decay, Do it for several velocities and note durations.
Now render this score to a wav file.
Then import that wav file, split it on silence and construct a soundfont from these samples.

The resulting soundfont thus could be made identical to the original one, but according to you should no longer honor its license?
No, the wording is exactly as it is to prevent such uses.

In reply to by jeetee

That's not at all what I'm talking about. My endgame is not to produce a new font. I'm talking about the poor shulmp (say me for example) who creates an mp3 to send to friend and both of us not be liable of some violation because of a lack including license info. Kind of silly for free software.

In reply to by bobjp

Which is why there is an ongoing effort to be able to include an actual free (CC0) soundfont in a future version. But until such a soundfont is created, we have to make do with what is available.

Again, this is just about the license of the Soundfont, not the MuseScore software itself.

In reply to by jeetee

I get this concern, but again, the distinction is pretty much universally understood in the music licensing world. (in the same way, I suppose, the MIT license is considered to be well understood in the context of the software licensing world).

When I say i's pretty much universally understood, I meant, to the point where I'd wager there has never been a single exception ever since the first sample-based synthesizers began to appear several decades ago. Without a single exception I have ever heard of, the intent of copyright on sample libraries is to protect against people making usable copies of the sample library, while leaving no restrictions whatsoever on the normal usage. Each synthesizer and sample library has its own variation on that license statement, some more clearly worded than others, but the intent is so obvious that no lawyer or judge would ever even consider quibbling about wording. They all come down to precisely the same thing: you can use this without restriction in the way it was meant to be used, just don't distribute anything that would actually compete directly with this and thus diminish the market.

That idea of diminishing the market, BTW, is a central key point made time and time again in copyright cases, in particular in regard to music licensing. Producing and distributing your music using a synthesizer or sample library diminishes the market for that synthesizer or sample library not one iota. The other relevant issue to consider is whether an infringement is considered "willful" or not. It could certainly be argued that MuseScore as a company is at this point aware of the issue, and could thus in some sense be on the hook if the original copyright owner(s) decided to sue. That's why I totally support seeing MuseScore replace it with a better-licensed soundfont. But the user who produces his own music would have no reason whatsoever to have guessed that MuseScore's soundfont would somehow be the one soundfont in the entire known universe that comes with such a restriction.

Not to mention, music and sound recording copyrights have their own unique qualities that would need to be considered here: the idea of non-digital copies, or broadcast, for instance. It would unclear how to apply the terms of the MIT license to that - you'd have to more seriously ask what constitutes a "copy" than would ever come up in the software world. The MIT license is indeed a poor choice for a soundfont in many ways, but again, judges are not (generally) stupid.

So again, bottom line, I think there is almost no chance that a judge who has even the slightest passing familiarity with music licensing and its long history and how copyrights have been applied with respect to sampling would look at that MIT license and say, "gee, even though every single other synthesizer manufacturer and sample library producer in the world has had the exact same understanding of how to license these, and even though the intent of the MIT license is obviously to enable software to be distributed for free and is thus less restrictive than a traditional software license, I'm going to nonetheless choose to believe that the person who assigned this license actually in this case intended to be more restrive than anyone else in the entire history of the music industry ever has been, and find Joe User to be in violation of the license for not attaching a copy of the license to his MP3 file". This to me is ludicrous.

And yet, all that said, sure, for all the reason state throughout this thread, it's a good idea for MuseScore to replace the soundfont with a one that can be licensed as CC0. These ideas are in no way at odds. One can say, "MuseScore should use CC0 soundfonts" and also "there is a virtually zero chance that any user of FluidR3 or a derivative of it has anything whatsoever to worry about".

In reply to by Marc Sabatella

Marc, I agree. Why would anyone care about how an audio file was made? But apparently somebody does.

But I'm not talking about just audio files. There's also digital artwork, videos, documents, and all matter of things produced by software to which no one attaches a license. Nor should they. You and I have clashed on copyright before, but it has always been on small details. As is this, I know. I'm just trying to look at as many angles as I can in an attempt understand what is going on.

In reply to by jeetee

I think one of the problems is that the licence uses the term “software” for the covered work. It should use “work”, and the licence I drafted, which is mostly equivalent to the MIT licence (but slightly better suited to match European law), does precisely that.

But in computing, when you actually have software, the effect is the same:

compiler (software) combines program source code (software) + libraries (software) to binary (software)

Here, the licence of the binary is, obviously, dependent on both source code and libraries. The licence of the compiler normally does not influence it (a famous exception is that some proprietary desktop and graphics software prevent you from using the documents you created with their gratis/low-cost editions freely).

In this example, the source code is the musical score, and the library is the soundfont. The soundfont is comprised mostly of samples (musical waveform), plus some processing instructions. The legal restrictions come from the samples to over 90%, the processing instructions are relatively minor but of course also contribute. But it’s the samples that are copied, initially verbatim, to the binary (WAV, MP3, OGG Vorbis, FLAC).

@jeetee you’d be surprised… you only need one to two dozen notes in one velocity for most instruments. The samples are used for multiple pitches; the Fluid (R3) Mono ones mostly have samples for C, D♯, F♯, A in each octave and often don’t even go to the extremes of the instrument’s actual range.

In reply to by bobjp

Please read again:

The software (the equivalent of your wordprocessor) is MuseScore.

The soundfont would be the equivalent of a clipart library, only that you not take one but several dozens out of it for a document. Sure you have to honour the licence of the clipart library for the final document.

In reply to by mirabilos

Interesting. I always thought of the term "derivative" in terms of authorship of the musical work but not the sound. I had not thought of the raw nature and quality of the sounds as a potential component of a derivative work. I guess I will stick to PDF. (that MAY be safe) Yet almost ANYTHING that becomes commercially big enough would have the eyes of the unions upon it in some fashion anyway.

In reply to by ramblinj

Fonts in PDF is another topic altogether.

But many fonts come under an embeddable licence (SIL OFLv1.1, e.g. Bravura, Gootville, MScore Text, MScore BC, MScore Tab, MuseJazz*, Campania) or a licence with a suitable font embedding option (GPLv3 with Font exception, e.g. FreeSans, FreeSerif, MScore) that this is not as much a problem.

In reply to by mirabilos

Sorry. "Copyright always applies" doesn't mean what you think it does. What the agreement says is that you are free to use clip art for personal use without restriction. You only have to include the licence info if you are offering the work for sale. You do not seem to be a lawyer, but this is not difficult to understand. But why do I bother....sorry, I couldn't resist. If it was required to include licence information in all cases. the agreement would say so. It specifically only lists the case of offering the work for sale.

In reply to by mirabilos

Whether something is offered for sale or not has almost nothing to do with copyright law - the relevant distinction is whether it is "published" in the sense of being distributed to the public. That is, distributing pirated copies of your favorite book/movie/CD/score for free is not any more legal than charging people for them. It's actually worse in an important legal sense, as it diminishes the market more than charging for it it would, and that's a consideration in determining "fair use" in the US.

Okay, can someone summerize EVERYTHING that was discussed?

Is it okay for people to use MuseScore's soundfont to compose their own songs without any copyright issues?

In reply to by Jojo-Schmitz

To update this question/answer, if possible, could someone post the best statement to use in attributing the work obtained from Musescore, when it is used in developing a sound file for commercial use? Thanks.

Do you still have an unanswered question? Please log in first to post your question.