"Nesting" of compositions


I think @Ludwig had mentioned in another thread the need to somehow tackle what he called IIRC nesting of compositions.

Here’s a bad example…

all 12 concerti of L’estro armonico combined into one composition. I don’t think it should be done like this, but what can you do with 3rd party metadata where they may have certain rules?

I can, of course “de-identify” the set and have my works from tags, but it’s a real dilemma, because I’d loose all the other metadata with it…

I think this should be tackled one way or the other soon. Beethoven’s op.18 come to my mind as well, etc.


1 Like

I posted the opposite problem, elsewhere: the 46 chorale preludes of Bach’s “Orgelbuechlein” each as separate “compositions”, some under a minute long. Part of the problem is the catalogs. Schmieder gave each chorale prelude, as well as each prelude and each fugue of the Well Tempered Clavier, a catalog number (though in the latter case it does, properly in my opinion, come out as single “compositions”, Book I and Book II), same as the Saint Matthew Passion or the Mass in B Minor–one number each. Beethoven himself published his first string quartets as his opus 18. I think Vivaldi was also the victim of external catalogers though I’m not that familiar with his works.

I think the best solution would be to give the user control over grouping and ungrouping “compositions” or “works” as they used to be called.

1 Like

I fully agree with your solution: give the user full control over compositions. In the “Metadata Preference” editor tab, add an entry “Work (Composition) Prefer Roon | Prefer File” similar to the ones already there for “Album Title”, “Album Artist”, etc. The “Work (Composition) Prefer File” setting would indicate that the WORK tag settings should always be obeyed.


When I run into this problem I edit the “groupings” tag in JRiver and set the disc# just in case. This has worked so far.

I never knew what the “groupings” field was for. I just enabled it in JRiver. They’re all blank. What do I put in them?

This is what I would place in groupings for Mahler’s First Symphony:

Mahler: Symphony No. 1 in D

This would be for all four (or five) tracks of the work.


but that would be contrary to the documentation in Roon knowledge base. According to that it is the WORK and PART tag that are used for this. And based on my experience, as soon as the album is identified, the 3rd party metadata takes over anyway.

would be interesting to hear a comment from the Roon team on this. @joel ?

I hear what you are saying, but Roon nesting did change after I changed tags in JRiver. I am not privy to Roon code, but I suspect their algorithm uses multiple sources for their results.

well, it’s worth a try for sure. I will report.


tested it, but for identified albums (and this was my topic) I observe no impact on composition grouping neither via GROUPING nor via WORK tag. That’s exactly what I would expect based on the documentation in the Roon Knowledge Base.

I was cleaning up my Roon compositions the last couple of days and this made me think again about this nesting stuff…

I think one of the big challenges for the Roon team when providing “automatically groomed metadata” is consistency in content.

My understanding at the moment is that several data sources are used by Roon, but I’m not sure whether the inconsistency I observe is because of different data providers or because of inconsistency among a single data provider.

I was cleaning up my Brahms piano works and there are a lot of “Klavierstücke” or pieces for piano that are collections of short pieces, covered with a single opus number.

Let’s take the 7 Fantasias op.116. For these I found a collective entry (Fantasias (7) for piano, op.116) as well as single entries for single pieces. Same is true for most of the song cycles.
This clutters my composition view for composers. In terms of tags the stuff is maintained consistently but not when I look at the metadata I get automatically.

Of course you can discuss how these “entities” should be represented, i.e. as parts of an opus collection or as single pieces with op.xx no.yy.

But please be consistent! If Roon wants to provide content (even via a third party) this needs to be tackled somehow. I’m sure the Roon team is aware of that but I’d be interested in their and other user’s view on this.

I’m quite sceptical that consistent metadata can be achieved without first defining the guidelines for its maintenance and having control over it. But how can this be achieved if the content is not “owned” by Roon? I don’t see a scenario where this can be solved by the software itself, i.e. based on all the magic stuff that Roon does now.

I could drown Roon support with “support tickets” for inconsistent or incorrect metadata - but where would it lead to? Would it get any better or would I just block resources that could do other or more important stuff? Just not sure whether it would be helpful or not.

What are your thoughts?

Here’s one angle on this:


thanks for the link. Crowdsourcing is something which is raised in these discussions all the time. It sounds nice on paper but it is a pain in the butt if you like to achieve consistency in entries.
I do not see how it could solve the “opus collection” vs. “opus + number” dilemma, for example. There is the need of some strict rules and guidelines that users need to adhere to, not all of which can be forced by the software itself. So there needs to be something like an approval process…
I’m afraid otherwise we would end up with a bigger mess than now.

For classical you would need to decide, what defines the work entity:

For Brahms op.116: is it the opus # which consists of 7 individual pieces (normally equal to tracks) or is each piece s single work entity?
For Beethovens op.18: is it the collection of 6 string quartets? If yes, how are the single string quartets handled? These are also multi-track individual “works”.

Personally I would be ok with Brahms op.116 to be seen as one “work” whereas I would not like to have it for Beethovens op.18… And I’m sure other classical loving users would have a different opinion/approach on this.

At the end there needs to be someone who sets the standard for this and I’d rather have a consistent standard to which I do not agree completely than a mix of user-defined “standards”.

Just look at the hits you currently get from freedb for classical

This is clearly an issue. There are several questions at large here. I think it’s crucial that Roon is consistent about this, not by making an arbitrary decision whether to display op. 18 or op. 18 no. 1 as the primary level, but providing an interface which provides linkage between the two, and remembers your decision about each page (an easy thing to do, which happens in other places in the interface).

So you would have a Work Details page for Beethoven op. 18, and Work Details pages for each of the 6 quartets (with the Rovi texts for each of the 7 items). And it would be clear on the pages that you can link from one to the other.

And on the Album details page you could have a right-click on the tracks which says or or something like that. That’s basically an edit of metadata the format of which we disagree with.

I don’t think crowdsourcing is right for this. I think Roon should mostly be able to see the connection between op. 18 and op. 18/1. Obviously there may be tricky cases, but I hope it would be doable.

Team Roon this is exactly what I am talking about in #16854.

1 Like

We’ve fallen into the trap of trying to discern simple and consistent rules for parsing composer catalogs before–our current system is significantly more manual and labor intensive. Machines still involved, but they need to know about specific composers, their catalogs, and the (many) exceptions. There is absolutely no way this sort of thing could be done automatically. The tricky cases are the rule, not the exception.

The music history “industry” is awful at making and sticking to conventions across composers and eras, even through eras where the idea of international standards and reasonably modern practices for handling data were well established (for instance–one of the catalogs most rife with invented inconsistency–the Hoboken catalog–was conceived in the mid-20th century).

I’d love it if the music history world would “grow up” and ratify a proper international standard, and pay the editorial burden of re-numbering everything under a system that could be understood by machines…but I won’t hold out my hopes.

In this case, there is literally no good source on earth for consistent data. Just a series of bad editors–starting with the people who assigned the catalogs haphazardly in the first place, then the people who printed potentially incomplete or erroneous catalog#s onto CD packaging, then the people at Rovi/Tivo who entered the data.

I place most of the blame on the music historians–if they had taken the care to create and document (originally or retroactively!) a crisp, mechanically verifiable, self-consistent, and comprehensible system that worked the same for every composer and reflected these hierarchies clearly, then there would be no way to get it wrong, and downstream users of the data could be reasonably expected to handle it properly.

By making things so complex/inconsistent, they have created a situation wherein only experts know how to deal with the catalogs + non-experts are forced to view the numbers as opaque, and also a situation where the amount of effort required to handle everything properly is too large to repeat in each system that might want to do so.

Because catalog#s are only comprehensible to experts, the only real solution is one where experts lay hands on the data and organize it properly. Whether that’s a centralized process–via a standardization process + renumbering, or ad-hoc–in a crowd-sourcing system like the one we may eventually create–experts must do the work.

Sorry, been holding that rant in for a long time. It is incredibly frustrating how “close but far” the world is from having a firm handle on this stuff, and also frustrating how simple people perceive this problem to be from the outside.


I agree. I don’t think any algorithm could solve that.

They don’t stick to conventions because there are none and I doubt there ever will be (at least not in my lifetime). None of these catalogues have been designed with IT in mind.[quote=“brian, post:15, topic:21546”]
Because catalog#s are only comprehensible to experts, the only real solution is one where experts lay hands on the data and organize it properly. Whether that’s a centralized process–via a standardization process + renumbering, or ad-hoc–in a crowd-sourcing system like the one we may eventually create–experts must do the work.

I have seen the most fierce discussion of experts on the internet about the numbering of the Schubert Symphonies… The “neue Schubert Ausgabe” has the unfinished symphony as No.7 and the great C major as No.8 instead of 8 and 9.

The danger here is that every classical music nerd thinks he is an expert…:grimacing:[quote=“brian, post:15, topic:21546”]
Sorry, been holding that rant in for a long time. It is incredibly frustrating how “close but far” the world is from having a firm handle on this stuff, and also frustrating how simple people perceive this problem to be from the outside.

I feel your frustration. I never thought it to be simple. That’s the reason why I just accepted in the past that I had to maintain and organize my works in a way that worked for me. I also have no solution for it and personally I would not have expected Roon to solve it.
I just need the tools, i.e. a solid data model with some clever maintenance screens… :wink: or at least some means to overrule the experts :wink:

I have lots of respect for you guys to even try to tackle it!

At least y’all can argue with each other instead of with us about it :slight_smile:

1 Like

oh no please don’t leave us alone amongst ourselves…:sweat:

1 Like

Even worse is that even real classical music experts think they are also therefore experts at programming… Isn’t that right, Brian? :wink:

I have no idea how possible any of the opus number stuff is, but it’s already awesome as it stands. Still, I’m sure you don’t want your classical fans to just keep their ideas to themselves, do you? So in that spirit we will carry on making probably impossible suggestions and I’m sure you will pick the best and most possible ideas and make something wonderful with it.

Thanks @brian

1 Like


I could not have said it better!

1 Like