"For" tag doesn't function correctly

• Nov 26, 2017 - 18:58
Type
musescore.org
Severity
S4 - Minor
Status
closed
Project

Whenever you visit a score on the website which has a few unknown instruments, the "For" tag underneath the title tells you that the score is "For [instruments], Harpsichord, Accordion and Percussion". See added uploads.

Attachment Size
bug2.png 1.1 MB

Comments

In reply to by jeetee

Status (old) fixed active
Status fixed active

Not fixed here (or not yet)
Before the fix, my scores were displayed "For keyboard"; now it's for "Guitar and Bass". Better, we get closer, but none of my scores are written for Bass

@Cadiz we group "Guitar and Bass", so it's basically one category. There is no single Guitar category. Here are the categories we defined:

  $instrumentation = array(
    '0' => t('Piano and Keyboard'),
    '1' => t('Percussion'),
    '2' => t('Guitar and Bass'),
    '3' => t('Strings'),
    '4' => t('Vocal and Choral'),
    '5' => t('Brass'),
    '6' => t('Woodwind'),
    '7' => t('Folk Instruments'),
  );
Status (old) active fixed
Status active fixed

Okay, seems unexpected at first glance. But we will get used to that, surely!

EDIT: A bass IS a guitar. So, maybe "Guitars" better?
(We already have a plural for "Folk instruments"). And for Strings.

The whole "for Tag" feature is counterproductive, as far as I am concerned.

First, a seeming bug. Why does this solo organ piece https://musescore.com/user/1831606/scores/3889661, whose instrumentation in the score is correct, "Pipe Organ", say "For piano and keyboard"? There is no piano mentioned anywhere in it. And is Bach's Royal Instrument no more than "keyboard"? What is the point of saying "Keyboard" if an organist is looking for music? If you're looking at a score, isn't the intended instrumentation obvious from it 100% of the time?

The whole idea of inferring and fronting prominent score metadata which is not under control of the user seems unprofessional. More often than not (no, I have no data, but I speak for everyone I know) MuseScore composers and arrangers use the sounds of MS instruments other than the ones asked for by the score because of the inadequacy of the "MS Orchestra" to produce the needed varieties of sounds. For instance, by using ingenious combinations of MS Sounds, @ClarinPardo has credibly simulated small pipe organ sounds with hidden masses of carefully-selected woodwinds, but his scores are intended for "organ", nothing other. People use hidden tracks all the time to provide execution details; this is not scoring for multiple instruments.

Novices and youngsters might create masterpieces for "kazoo and guitar" knowing little about issues with either instrument, and the MS instrumentation can reflect their intent exactly. When I prepare a score of a Bach aria for Soprano and Continuo, which latter I implement with Pan Flute, Cello, and contrabass, I want it to say "Soprano and Continuo", not "Voice, Woodwind, strings, Piano and Keyboard." Bach did not write for Pan Flute, but the sound of MS Pan Flute makes a great continuo right-hand (real Pan Flutes can't even play chords, let alone counterpoint). As shown above, trying to detail large ensembles (e.g., "symphony orchestra with some rock instruments") is pointless.

"String quartet", "chamber chorus", "Soprano, chorus, and organ", "2 violas", "string quartet and trumpet", "violin and continuo" (or "double SATB chorus, boy's chorus, double orchestras of strings, 2 flutes, 2 oboes, and continuo" :) in a summary listing might really help someone choose or categorize pieces, but you can't ascertain such categories without user input. When one has nothing accurate or useful to say, stay silent.

This new "feature" (who asked for it?) is for youngsters and dabblers, not experienced, knowledgeable MS “power users”. What can I say to get rid of it or have a way to? Whom does it help? Whom will it mislead?

Even "140 characters" has become "280" of late.

I see 2 different ways of doing it right: a) take the instruments name straight from the score, from the staff properties and b) take the GM sound assigned to the instruments. The former should be the most correct, the latter might be easier to obtain (it is taken for the playback already)

In reply to by Jojo-Schmitz

How about "Let the user say what forces he or she intends"? You sound as though the user has to carefully read the midi/staff properties and take the instruments off a digital shelf and plug them in before playback. Hasn't the score author done that already?

Why is "correct" not equal to "what the user wants"? Are you saying that using "instruments" other than the ones with the same names as the one you want is a non-supported misuse? If I want "organ", it has to sound like the single MS "Pipe Organ" registration, or I'm abusing the software?

Hi? Taking the instrument out of the staff properties is what the user picked, even if for playback purposes (s)he later picked a different sound.
Currently though muscescore.com seems to pick on the sound and match that into a category.

What was wrong with the old way when it listed the instruments in the score? Some of the lists got long, but it was easy to find scores for a tenor sax if you wanted one.

In reply to by Jojo-Schmitz

Did you not read my long post two above? Is the argument that the MS sounds used, often hidden, do not necessarily correspond to the instrumentation the author is writing for, or the argument that enumerating instruments is only in the simplest cases the most useful or helpful description of required forces wthout appeal or power for you?

If the score is intended for MuseScore performance, then the intended instrument is not flute, chorus, organ, or orchestra, but "MuseScore", and everyone has one available. Listing the midi instruments used is as superfluous and useless as listing the bytes of the file.

But if the score is intended for human forces, listing midi instruments used in any capacity, no matter how divined, especially hidden ones, is incorrect in enough cases as to make this information embarrassing for serious authors. Why is this so hard if the visible product can be made better by simply removing this often-incorrect new display?

I can control what the score looks like, and with enough devious ingenuity, what it sounds like. Why can't I control how it's summarized or indexed? Why MUST this be a magic, tone-deaf formula based upon the sounds I used to get the computer sounds I want for visitors?

If you're telling me that I and others "should not" be writing scores where the intended instrumentation is not that chosen from the MS Midi palette, please explain why you consider that to be within your purview. The intended instrumentation of a score is wholly the choice of the composer, to be conveyed to performers and audiences. Web sites and computer applications have no role in that choice. The computer details (e.g., midi instruments, hidden tracks) used by a MuseScore author to make a web-site copy of that score (whether his or her own work or that of a famous composer) "play" properly are not at all the same thing. Period.

I'm brewing on a solution. If only 5 instruments are used, I'll use the exact instrument name. Not the MIDI one, but the instrumentId from the MusicXML ontology. For more, I will use the instrumentation naming.

Just so it's clear, outputting the instrument names is useful for visitors to be able to quickly see for which instruments the score was created for.

In reply to by Thomas

What do you mean? How can the user control the "Music XML Ontology"? What problem does this solve? How does this help anyone who does not use the same MS instruments as he or she intends to score to be performed with by humans? Are you saying or are you not saying that you have no sympathy for that extremely common case, or it is an abuse of the software? It's as though you think that this need is so unreasonable that no response is needed. Please explain.

You say "users can quickly see what instruments the score was created for." If I the score is created for organ, but I am forced to use various recorders and brass instruments to get a sound that roughly approximates the organ registration I need, the "score was created for" organ, not recorders and brass instruments. Am I being unclear, or do you not consider this a legitimate need?

In reply to by [DELETED] 1831606

I would say your workaround case is less common that simply using the instruments listed in the instruments list. As far as changing the sound is concerned, you can use any instrument name you want with any sound by using the mixer. If you want the score with workarounds only listed under organ, then use 5 organs and change the sound in the mixer on the other 4. Too many staves you might argue? Delete the unneeded staves in the instruments panel. That way you are causing less confusion for someone looking for an Organ solo score.

As for the Continuo, that is not an instrument, it is an indefinite group of instruments. As in the case of an actual performance back in the Baroque period, you decide on the instrumentation when you transcribe the piece. Unfortunately, There is no Continuo sound (perhaps made up of a Harpsichord, Bassoon and Double Bass playing in octaves as appropriate for example) that can be used or I would suggest adding a Continuo instrument to the MuseScore instruments list.

If I write a score for SATB, I want musescore.xom to show either vocals or Choir or Choral, or, correctly, Soprano, Alto, Tenor and Bass, regardless whether I for playback purposes picked the sound of flute, violin, oboe and trombone.
I want others to see which instrument this score is intended for, not what sounds I used for whatever reasons.

And Thomas' proposal is the same, so we're all in agreement.
This does negatively impact older 1.x scores that used violin or trumpet as the instrument and different sound to emulate an instrument change though.

In reply to by mike320

No need to explain to me what "continuo" is: see my manual on continuo realization, https://musescore.com/user/1831606/scores/1745621 , which has a whole section on MuseScore issues. Continuo is indeed not an instrument, but an understood (by competent musicians) performance force. A Baroque sonata written for violin and continuo is just that, not violin and cello, not violin, cello, contrabass, and harpsichord, nor violin, pan flute, "piano and keyboards". "Violin and continuo" is how it should be listed, and if you have no idea what "continuo" means, you shouldn't be attempting to perform it with humans and instruments.

Like a living ensemble-director, the MuseScorer must figure out how to realize (both in instruments and harmonies) notated continuo. That is not the same as rescoring it for the chosen instruments.

Soprano is an instrument you add via the add instruments dialog, it gets recorded in the score with a MusicXML ID of voice.soprano or some such, this is what musescore.com should use.

In reply to by Jojo-Schmitz

Are you saying that the instruments that you want to appear as "for" must be those and only those added by the Instruments dialog, and any other re-instrumentation you wish should be done with "Instrument change" or Mixer manipulation? It's not a total solution (all posted scores with "implemetation instrumentation" will have to be rescored and reposted). This doesn't hidden tracks, either.

Instrument change should be part of the For, if possible, mixer changes should not. Bad luck for continiuo I guess, just pick whatever instrument you want to play that continuo, it got to be some instrument in the end anyway.
Workaround might be an instrument of the name continuo, but without an Xml Id, so musescore.com falls back to use the name rather than the Id. Such a fallback is needed anyway.

This discussion falls under the challenge to create a better categorisation on MuseScore.com. The first and most simplistic way we followed was through the midi numbers. Since MuseScore 2 we are now also retrieving the instrument id in the MuseScore file which is based on the MusicXML ontology.

Leveraging this information, along with user input after upload, we should be able to get to a better categorisation, using the naming people typically use such as SATB, Marching Band, and more.

Concretely, can we together come up with a list of instrumentation names which are commonly used by musicians? Along with these names, also the list of instruments (instrument id as in musicxml) which commonly occur in each instrumentation.

In reply to by Jojo-Schmitz

You could add Continuo to the list of instruments and ask what the preferred default sound would be. This would allow the .com site to be able to list it. Of course this wouldn't be fixed in MuseScore until the next release.

I think this would be a very reasonable addition since so many Baroque scores list it as such.

@Thomas:
Simply use "Long Insrument Name" pls.
Using xml-id or other midi-names isn't correct.

and Trust to User.

If there is a staff I named as "hand saw", I want to search for it in the same way. not as "synth.effects.goblins" (xml-Id) or "Goblin" (GM name)

Hold on a second, Continuo is not only not an instrument, it isn't even a staff of its own, but just some numbers added to another staff, so doesn't even play back at all. Realization via some hidden staff is entirely different, and I don't think hidden staves should get considered by that For list.

The solution lays in doing a smart combination of the information we have, along with the information the author can further provide after upload, all while not just dumping this information on the website, be as consice and useful as possible.

In reply to by Thomas

Thomas' last message (intelligent combination of what is there with USER INPUT AT UPLOAD TIME) agrees with me greatly. In the meantime, PLEASE get rid of the new "for" on the score pages.

About Continuo, that is not correct. Continuo is the sum of a variable number of bass instruments and a "realization", not one or the other. The single line with the bass and the figures appears as "continuo" in scores, and the MuseScorer or ensemble leader must provide for realizing it. The "realization" does not appear in academic/library/"complete works" scores, but modern performing editions often include a realization, so you may or may not want your realization (or someone else's that you copied without paying) to appear, in regular-sized notes or small. All I really want is to be able to label a staff "continuo", another "realization" (which I can do today) and NOT have the score advertised as being "for" the instruments I have chosen to be the continuo.

The problem here is not the application, the sounds, or the scores, but this crazy "for" feature which exploits information to which I do not think it is entitled with unmediated access.

Long Instrument name seemed like a pretty good guess, but that fails when instruments/vocals play "roles", e.g., "Rigoletto", "Evangelist", "Violin 2", and Long Instrument Name gets that. But that's clearly getting warmer.

Partname is maybe what is needed here, i.e. the instrument picked from the add nstruments dialog, has also the advantage to be translated into the user's language.

In reply to by Jojo-Schmitz

I suggested a solution that makes sense to me, with the rationale of "When you appeal to something, you have to offer something else that can be used instead"
The reason is that when the user changes the name at the head of the Staff(1), this change appears in the "Long Instrument Name" section.


(1) With double-click or right-click on it and using "Edit Element" command.

OT (or not OT?) I can't find an old (some years ago) conversation.
Someone didn't see the long instrument name on the scores loaded on MsS.com and he had found a trick, Some changes on the staff?
But I don't remember more and probably it is not relevant, only curiosity if someone remembers better.
(Please excuse me if I am inopportune)

It is exactly what is needed here, 100% exact and under the user's control. You can put your continuo there

Hmm, maybe not, that list is just too long for e.g. a full orchestra, there we'd rather want to see For Orchestra, or For Bigband or For Strings Quartett, For Brass Band, For Marching Percussion etc.

In reply to by Jojo-Schmitz

The page organization here leaves a lot to be desired. I can't tell WHAT you are saying is exactly what is needed here. "Part Name" is not an instrumental requirement: "Violin 1" is not the name of an instrument. "Long Instrument Name" is equally poor: the score should say the name of the operatic character etc, not "tenor".

Yes, for large orchestra we'd like to see "symphony orchestra". "Big Band", "string quartet", "SSATB vocal quintet", "violin and continuo." I said all that hours ago in this thread. The obvious correct answer is to allow the user to SAY exactly what he or she wants in the "For" (if anything -- the score could be a tutorial or abstract work), with no reliance on midi instruments.

I had to leave for a couple of hours so pardon me for a late comment.

As far as following the conversation, the issue tracker is not designed for conversations like this but rather a linear record of testing bugs and comments about fixing it rather than the forums which use indentations to help follow conversations. But we are here, so we'll just have to read carefully to follow the conversations.

The problem with user definable instrument names is that most people would not be looking for a Figaro but a tenor voice, so IMHO none of those should be used in the instrument list.

I understand exactly what a continuo is. If I found continuo in the list of instruments I would expect to find a Baroque style score with a continuo line that probably includes figured bass. For notation purposes adding this as an "Instrument" makes perfect sense. (How often is it stated on this sight that MuseScore is a notation software). Playback of the continuo probably won't matter to a lot of people.

@Thomas said, "The solution lays in doing a smart combination of the information we have" Perhaps some sort of user defined tags for the instruments would be a way of doing this. The .com site could create a list of tags based upon the instruments it finds in the score and allow the user the have the option to edit the list. This would allow the early example of the Organ with workarounds to have only the Organ listed.

Now that I understand why my organ score says “Piano and Keyboards”, i.e. that’s the “group” containing “organ”, may I note how unsuitable this for visitors (i.e., not desktop app users) who could reasonably be expected to read “Piano and Keyboards” as an indication that at very least a piano, and perhaps other “keyboards”, are required for the work. A work for solo cello (of which I have 2 posted) is not for “strings”, although a string quartet would be. I hope I am making it clear that (1) the “group” business is, in general, not the right thing for “for”, and (2) “for” cannot be done without user input, and (3) that this page feature should be rescinded until there is sufficient infrastructure to support it usefully. If, as @Thomas has told me, a goal is to make the site intuitively usable with minimal or no documentation, then labelling all organ or harpsichord (or piano, even) scores as “piano and other keyboards" fails that goal madly.

Another takeaway for me from yesterday's discussion is the lack of a staff property "role"/"label", i.e., "Figaro", not "Baritone", in short and long variants. "Part Name" clearly is not this (that should be called "part tag", I suspect), nor is "instrument name". When a part has a "role", be it "Figaro" or "Violin 1", that is not an "instrument" which can be searched for or by which the score can be indexed/categorized, but exactly what should appear (in long or short form as appropriate) on the visible score. At least for vocals, this is a need in every opera, oratorio, or "show" score. WRT instruments, "violin solo" or "flute 1" are not the same as "violin" and "flute". While I appreciate compatibility issues, it seems as this area of MuseScore (application) could benefit from careful redesign.

But "Figaro" and "Violine 1" is an excellent candidate for the Long Instrumentname, which is what is shown in the score. Partname is more sort of meta data, data about the score, Instrument is too, but additionally determines sound, clef, range, transpositioning

Thank you for all your feedback. I conclude that this feature needs work, so I have hid the instrumentation output on the score page.

The solution which I'm working on will be based on Jojo's idea to re-use what we have already in the template taxonomy at https://github.com/musescore/MuseScore/tree/master/share/templates

Concrete:
* introduce new instrumentation field on the online score form
* the values will be a discrete 2-level list from which the user can pick only one value
* a suggestion can be made by the system based on the instrumentId information retrieved from the uploaded score

Values:
https://docs.google.com/spreadsheets/d/1Xtuynf2g7_9s9Yo_fZHhCJDu0WV0teP…

Thanks for disabling it until it's ready. I don't understand your proposed solution: that list of common ensembles omits all duos and trios, e.g., "Piano, violin, horn" (Brahms op. 40), choral ensembles not SATB, various combinations of orchestral resources and vocalists, etc., etc. Please show us some examples of what this would look like for typical and atypical classical and non-classical groupings. There don't seem to be enough shoe sizes here, nor will any pre-built roster have enough.

It goes without saying that people here often take joy in attempting movements from the grandest scale works, operas, Mahler/Bruckner/etc symphonies, grand masses and Passions, in which modern composers do not hesitate to include exotic instruments (including rock) among huge ensembles. How will these be usefully represented? The St. Matthew Passion of Bach requires two choirs, orchestras, and sets of soloists, but his St. John only one. Yes, vast movements from both are already represented here. What will happen in these cases? How will a user cooperate with the website to produce a meaningful description for large (well-known, beloved) works with unique resources?

In reply to by Thomas

I'm more interested in the topology of the schema here than in listing all the ensembles I've ever heard of (and I am not encyclopedic!). If I were to attempt to describe this mathematically, I might take your spreadsheet, augment it by all the single instruments we know, so it contains not only "rock band", "symphony orchestra", "strings", and "barbershop quartet", and CONTINUO!, and then say that a "requirement list" is a list of any number of those, each with a how-many. I.e., "this is a piece for 2 SATB choirs, tenor, band-saw, rock guitar, and string orchestra". That'd be a START. (of course, you need a "language" of choirs, e.g., Domenico Scarlatti "Stabat Mater for SSSSAATTBB and continuo" etc.). Brahms' op 40 trio ("horn, piano, violin") is not a "grouping you have to support", but should fall out naturally because you support horn, violin, and piano (the language of the extant system handles this ok, but does not handle understood "ensembles" (e.g., "rock band")).

The next topological level would be multiplication of arbitrary groups at arbitrary levels: the opening Chorus of Bach's St. Matthew is for TWO orchestras of strings, 2 oboi, 2 flauti and continuo, TWO choruses SATB, and ONE chorus of (boy) sopranos.

Of course, this is exceedingly complicated to represent, explain, and search. But that's the actual way ensembles are organized as I have experienced them-- by hierarchical groupings with meaning to the composer, interpreters, and listener. Your task is where to choose the tradeoff between this full complexity and listing midi voices.

In reply to by [DELETED] 1831606

Another issue is "what is the intent of the index/search facility?" Accurately representing the required forces for a large work is a different problem from characterizing the forces required for a work in a way useful for different kinds of people searching for say, music they or their small group can perform. For example, for an oboist looking for "oboe music" that she and her pianist friend can play, it really doesn't matter how you describe large symphonic/choral movements -- "large symphonic/choral movement" is good enough, she can't play it, but "oboe, flute, and piano" would be interesting to her, as would "2 oboes and organ" if she didn't have an oboist friend. So the amount of detail required at various levels would depend upon the intended use. The SSSSAATTBB + continuo of Scarlatti's Stabat Mater could be "large choir and organ" to first approximation. But if you say SATB, you are suggesting that four people can sing it (even if written for chorus). So there are numerous human-utility issues in addition to the aforementioned "representational power" issues.

Here's a theorem from those postulates. It is more useful to accurately detail soloists and small ensembles, and to characterize large ensembles. No one is going to look at these indexes or listings to find out exactly what you need to perform Mahler symphonic movements (of which there are plenty on MuseScore), but for small groups or soloists, they do want to know exactly. I don't know if the decision can be algorithmetized, and how to avoid ridiculous results. Proposals and experimentation are necessary, as well as a documentation/explanation strategy. How do you explain to someone who doesn't know what a "continuo" is why countless baroque movements call for "one"? Or why SSAATTBB is different from SATB SATB?

In reply to by [DELETED] 1831606

Here's another real-world requirement: the countless, beloved "songs" (Lieder, Chansons) of the 19th ct. and later are written for generic human voices, but each such song is either for "high voice" or "low voice", and transposed editions are regularly sold for whichever the song wasn't written for (and men and women have regularly performed them all). "Voice", "high voice" and "low voice" are thus "instruments" for which singers look (I am trying to recall all the now-vanished "music stores" in my metropolis). Similarly, there will be "scores" on MS written for no instrument at all, e.g., lessons in harmony, counterpoint, composition -- of which I have written and posted a good number now. Saying "for piano" or whatever "midi instrument" i chose because I had to choose one to make the sound is not useful to pianists, learners, or anyone else except a developer debugging the midi file. The "Well-Tempered Clavier" of Bach, like most (not all) "Clavier" works of the Baroque, was written for any kind of keyboard instrument you might have (e.g., harpsichord, clavichord, virginal), although it has become a staple of piano literature for which piano-specific editions abound. And works like the "Art of Fugue", which are not intended for any instrument or instruments in specific at all (and remains one of the imperishable monuments of all of humanity's musical heritage).

File under "Generic and unspecified instruments".

I hope my comments in this thread are helpful and not taxing the patience of subscribers. I really want these ideas here for the "record", esp. in light of @Thomas' seeming invitation. I don't know how many readers remember brick-and-mortar "Music stores", such as Briggs and Briggs of blessed memory here in Boston, and how they arranged "sheet music" (as even MuseScore used to call it). You can look at online music stores (I just visited sheetmusicplus.com), and see how they "file" music. There, there are four menu-level categories ("Choir and Vocal", "Piano", "Band and Orchesra", "Instrumental"), and tens of subdivisions under each; in this taxonomy, "voice and piano" is a subdivision under "Choir and voice". I don't even see "organ" on any of their menus (but I know they do sell organ music). You should look at other online music stores for their taxonomies.

All of this forces the question of "what is the purpose of the 'for'?" Is it to present as a "Music store", which customers visit looking for a particular score, or maybe even browse scores "for" (in the most general sense) some particular instrument in some capacity? The correctness of a proposed design of the "for" depends entirely upon the envisioned purpose, or purposes, and if there are many, how to balance them and trade them off.

Can a movement be of multiple "categories", as per multiple conceptualizations of it? Unlike the brick-and-mortar store, the "sheet music" doesn't have to reside in only one "place". I have posted many cantata aria movements. If it were paper, should a typical one be under "Cantata and oratorio movements", "Bach, J.S.", "soprano, oboe, and continuo"? The ability to enter a title into a site search seems to obviate many of these. To paraphrase a renowned politician, "It all depends on what this 'for' is for."

Here's another thought. How reasonable is the "music store" model for this site? I don't have a handle on numbers, but it seems to me that "accurate scores of established repertoire" do not comprise the majority of this site -- that's not its purpose. Amateur and tyro composers "just trying to compose something" posting fragmentary beginnings of abandoned masterpieces (If I had a dime for every one such I've written... but I don't post them) and the like, and the openness of the site (i.e. it is not curated or "edited", thank goodness) vitiate its potential utility as a "music store". How often does someone come here and say "let's see what there is 'for' ... " ", and what are they likely to find if they do, no matter how 'for' is defined? Small accurate-score projects ("Open score") notwithstanding, the business of this site is neither "library" nor "music store": it's "platform"/"venue". Searching for "what music is here?" is not useful. Searching to see if anyone has an accurate score of "Frukoswki Clarinet Trio", on the other hand, is a different (and reasonable) problem.

And, as I've said, cataloging a movement for large, varied, complexly-structured forces by the exact instruments and vocalists required (other than in the score itself) is not solving any reasonable need. No one who has access to, or responsibility for, large forces ever "browses a music store for music to play/sing."

And, again, as I've said, anyone who wants to hear a score uploaded to MuseScore, even video-backed scores, already has all the instruments (i.e., MuseScore) necessary.

In short, I don't think any "for" feature, or searchability by performing forces, is needed/wanted at all. In three years here, I've never needed its help. If I've really overlooked some important use case, please speak up.