Apologies if this topic has already been discussed (I was not able to find it in the forum). I was wondering if it would make sense to have a structured way to indicate the instrumentation for classical works.
Currently, this information sometimes appears as part of the title (e.g. “Sonata for Piano”) or can be inferred from the “Work Type” (e.g. “Symphony”) but there is no way to enter it in a structured and systematic way, and for many works, this information is not traditionally given as part of the title (e.g. “English Suites, BWV 806-811”).
While it is true that there exist works whose instrumentation was not fixed at the time of composition, it seems to me that there is an intended instrumentation in the overwhelming majority of cases. Perhaps surprisingly (to me at least), we have a way to indicate that a later work is an arrangement of a previous piece for a specific instrument, but seemingly no way to formally indicate the original instrumentation.
Note that this is also different from the recording’s instrumentation which can differ from the original intended instrumentation (and which we can already indicate as an attribute to a recording-artist relationship).
It seems to me that this could maybe be implemented as an attribute of the composer relationship, or directly as an instrument-work relationship. I’m still quite new to MusicBrainz, so I’m sure I’m failing to see the pitfalls of such a feature and I would be curious to hear the community’s thoughts about the pros and cons of this proposal.
Hello, @thibauth:
Welcome to the forum, and thank you for 10 years of contributions to MusicBrainz. I am glad to see other editors of classical music releases.
I agree, instrumentation is valid metadata about a Work (the composition, in the composer’s intentions) and about a performance (in the performers’ actual choices). However, it is a bit complex.
What do you do if the composer wrote a piece for an instrument which was common then but unknown now? What do you do about the difference between the composer’s choice of instrument, and its modern descendent? (Think of “viola da gamba” and “cello”.) Do you want to go so far as documenting what the composition of the symphony orchestra should be or is? Should we say enter many first volinists and second violinists are in the ideal orchestra for a work?
It might also be hard to collect. What sources would provide this information? If we are lucky, the publisher printed it on the score, and the editor can find a copy of the score, and read the text about the instrumentation. If we are not so lucky, it might be hard for editors to source this information.
What is the value proposition? A lot of what motivates me to add content to MusicBrainz is some idea of who will benefit. I enter the content which will improve the tags which Picard will put in my digital music files. I enter content to describe local classical musicians whom I want to support and document. I enter content like Works which I am confident will help other editors. What will instrumentation information be useful for?
I encourage you to start by collecting information and putting it in the Annotation field of Work entities and Release entities. In general, the Annotation is the fallback home for data which does not fit in the more structured fields. As you add it to the Annotation, figure out how to make your entries consistent and well-structured. This can become a prototype for a more formal database field proposal. Or, it can be good enough on its own to meet the immediate need.
As you make entries that you think are worth review, consider making further topics in this forum and link to them. That might help you find supporters and collaborators for instrumentation metadata.
Best wishes with this initiative! —Jim
5 Likes
Hi! We do want to add this - the reason we haven’t yet is because right now we could add it for instruments, but not vocals (so we could say a piece is for violin and piano, but not soprano voice and piano). See STYLE-470.
1 Like
A use case I can think of is “I want to find works for violin and piano by female Argentinean composers” or whatnot 
Might not be something people would use, dunno, but it seems like a fun way to try to find new pieces for one’s repertoire.
1 Like
Thanks a lot both for your answers. I completely agree that there are some edge cases and we will need to think about these carefully to be able to provide reasonable guidelines for how to use the relationship if it ends up being implemented. But there are edge cases for virtually every concept and relationship in MusicBrainz’s ontology, and I don’t think their existence undermines the usefulness of being able to provide the instrumentation for the overwhelming majority of the cases in which it is well-defined.
Echoing @reosarevok, I do think that it could allow for interesting search or filtering features in music library platforms/apps (many or most of which rely at least partially or indirectly on MusicBrainz’s metadata). This just my opinion, but instrumentation metadata feels at least as useful as key metadata. More generally, I interpret MusicBrainz’s mission as being “the ultimate source of music information” independently of any specific application. My experience is that making structured and high quality data available often unlocks new and interesting applications in unexpected ways.
Generally speaking, for classical pieces, the authoritative source for instrumentation is the score as published during the lifetime of the composer, or the manuscript if the piece was never published. This sometimes needs to be complemented by other written sources (for example, correspondence between the composer and the publisher). This is how catalogues (for example Kobylańska’s catalogue of Chopin’s work) establish instrumentation and these catalogues can thus also be valuable sources for MusicBrainz’s metadata. Instrumentation could be compared to the date of composition (which is currently implemented on MusicBrainz): both are established in similar ways and exhibit edge cases.
Again, the score is the authoritative source; the number of violonists is typically not indicated.
Wikidata has properties orchestration and instrumentation which serve a similar purpose, and IMSLP has a detailed page about how to indicate instrumentation. I would be happy to write a report about instrumentation data on both websites (how it is added, used, edge cases, etc.) to inform stylistic guidelines as the feature gets closer to being implemented. One could also consider sourcing information from both websites in a semi-automatic manner (the linking between Wikidata, IMSLP and MusicBrainz for classical works seems to be quite good overall).
1 Like