I’m planning to revise just this area of my Roon Library at the moment.
So far I’m leaning towards using Genres (and probably my own, rather than Roon’s own batch - little like @Mike_O_Neill 's because my Library is exclusively classical) and not Tags.
I may well use a period-based system, viz:
Mediaeval
Renaissance
Classical
Romantic
C20th/C20th
and avoid ‘named centuries’ with the exception of the last one. If I give C19th to Schubert and to Liszt, I’m being inconsistent. Classical to the former and Romantic to the latter seems more logical.
And then Genres for the likes of:
Chamber
Concerto
Instrumental
Opera
Renaissance polyphony
Symphonic
Vocal
etc
But I might also avoid mixing the two by allotting time periods to Tags after all.
In terms of sorting by Album title, I’ve yet to come up with a solution to displaying Albums in alphabetical-order-by-title where they are prefaced with a number. This is true of the BIS Suzuki Bach cantatas. After Volume 1 come Volumes 10, 11, 12 etc to 19; and only then Volume 2 etc.
This is a (VALID) very different way of approaching the subject. I rarely use Genre and never use the Period . I my manually tagged files it isn’t even included. I suppose I view the composer period as included eg Beethoven etc.
You can get around this by Zero prefix ie 01,02 etc then they sort properly. Text doesn’t sort well, numbers do
I sense that you are surprised at the levels of grooming you observe in the roon user base but surely it is an obvious consequence of roon’s design choices? I was very surprised to read that a grooming-free experience was a design goal. Frankly, my experience has been quite the opposite that roon is by far the player with the highest grooming requirements. That is all genres, not just Classical. I think that is just being hidden because I, like many others, find the value add in heavily grooming a non-Classical library just doesn’t add enough value from accepting roon defaults. That doesn’t mean that many of the issues found in Classical libraries do not exist in other libraries like Dance/Electronic. For example, roon doesn’t handle re-mixes properly and this has a lot in common with roon’s way of handling transcriptions in Classical which roon doesn’t do properly either.
Multi-part grouping is a very good example of why roon is very far from a grooming-free experience. Compliance with TiVo composition structures is simply not a requirement with other players. This is an entirely new grooming requirement imposed by roon. Quite a step change. Of course, it is a choice but if you want to take full advantage of a roon Classical library and if you want all compositions identified and grouped correctly then the grooming requirements are very steep. There are just too many automated identification failures by roon so high levels of manual intervention are inevitable. By far the easiest way to fix roon composition identification and grouping failures is to cut and paste the TiVo structures from allmusic. This is a regular recommendation on the forum which always comes as a surprise to new roon users so I do not believe that anyone is doing this because it is a preference. It is a consequence of roon design choices.
Personally, introducing a “better” composition structure than TiVo that could take account of all possible composition hierarchies would be welcomed by me. Especially if it improved automated identification accuracy and reduced the need for manual grooming. But by the sounds of things that is probably no longer a realistic option in the existing roon. It would just cause too much chaos at this stage unless there was some bullet proof migration/upgrade process. I just don’t know how realistic that is. It may be more realistic to draw a line under roon and introduce something entirely new although incompatible.
I have a large local classical library as well. I just do a search like “Beethoven 5 maazel” and Roon usually finds what I’m looking for. I haven’t really bothered grooming at all.
The way I look at it: if I have to spend 10 min on clicking on different things to groom every album I buy, I’d rather just take my chances and spend that 10 min on the 1 out of 10 times when the search fails to find it…
Isn’t that the point of all this technology? To not have to do all this manual work?
Or it could be that I’m just a lazy sod.
On a related note: does the Roon API provide an entry for an external search engine? If so, an ML-based engine could do a much better job of implementing search based on all the available training data.
And that is what I used to do with legacy players. With simple text and string based searches, other players often do a better job than roon which is at the root of a lot of the complaints about roon search. I would say that before roon I did almost no grooming at all because a text string search was good enough.
But roon works in a completely different way and in effect chops and dices the long strings you typically find in album and track title metadata (and still do) into discrete composer, composition and performer database items. The advantage is, that means in principle that not only can you list out all the Beethoven 5 performances you have and all the Maazel performances you have but you can also link them to the universe of other performances on a streaming service.
This you cannot do with other players but the rub is that because there are no meta-data standards, getting roon working at its potential in this way will require a great deal of manual intervention. Just to put some numbers around that. Probably because I had no real library system prior to migrating to roon 7 years a go I had to manually edit 100% of my files before they would be recognised in any sensible way in roon. I have not finished migrating and just gave up the impossible task years ago. Even now I would say I feel the need to edit 70% of new editions to my library. This can rise to 100% with some composers where roon historically has never been able to interpret the complexity of the Classical cataloguing conventions. Haydn, Telemman, Debussy and many living composers come to mind. There are many posts and change requests going back years now on this and related topics.
I see from regular comments on the forum that getting a roon Classical library working at its full potential like this is of very little interest to the vast majority of roon users, even Classical users. But this is the point of this thread which is to share experiences and propose labor-saving pointers for those that are.
tripleCrotchet, I’d be interested to understand what you mean by “This can rise to 100% with some composers where roon historically has never been able to interpret the complexity of the Classical cataloguing conventions. Haydn, Telemman (sic), Debussy and many living composers come to mind.”
I have a very large Classical collection (over 8000 albums), and I never seem to have any problems locating / browsing it. I DO very quickly check the tags via MP3Tag before adding to Roon, but only wind up actually editing maybe 20% of them, and usually only standardizing the composer name and the genre - everything else is usually close enough
Anyone remember the old SIGs (Special Interest Groups)?
Given the specific needs of classical music lovers, I wonder whether there would be any interest in, enthusiasm for and/or momentum behind the idea of some sort of Roon Classical Music User User Group.
Nothing formal or even vaguely defensive.
Aims (just my suggestions off the top of my head after three years or so of happy Roon use):
to encourage RoonLabs (and now Harman) to give equal resources to advancing advancing Roon’s features and functionality where - for a start - it’s Composer > Work > Artist, as opposed to Artist > Song or Collection
to aggregate as much material as possible as a resource for classical music lovers enabling them to make the best of Roon… FAQ, links, threads like this one, expertise etc
I’ve now thrown my lot in with Roon 100% - happily. And am now contemplating whether/how to rip my several thousand classical CDs (when, as those here know, meta tagging afterwards will take five times as long as the actual rips) or rely on Qobuz and - ideally, if Roon ever integrates with it - Presto Classical.
(I have a couple of dozen items in Audirvana - mostly for reasons of nostalgia from the sort of things I was listening to (pop, jazz) in the 1960s, comedy and speech plus recordings from YT etc. And whatever I’ve been unable to get out of iTunes/Apple Music.)
Just (my) ideas Happy to see them rejected - or improved upon.
If you have a large library and you are just accepting roon defaults without a lot of double-checking I suspect you have a lot more composition identification failures than you realise. This can be with very famous works. Of course it is possible that for some reason you have a personal cataloguing system that roon never makes mistakes with. My experience is different.
Let’s take Debussy. Lesure made two versions of his “L” number catalogue. One in 1997 and one in 2001.
To cover all bases, TiVo uses both 1997 and 2001 L numbers in its composition naming, creating the potential for all sorts of ambiguity and confusion. As I understand it TiVo uses the prefix “CD” to distinguish the earlier L number from the later L number and roon carries that over to give this sort of composition naming convention:
This is just a rather random example but unless you have gone to the trouble of disambiguating L. 82 from L. 75 you will find roon misidentifying famous Debussy compositions like this. The problem arises because your metadata may use just the older or just the newer L numbers (instead of both) and it is not clear to roon which you mean. Personally I have found Debussy composition miss-dentifications to be so frequent that roon has trained me to copy and paste allmusic meta-data on any Debussy album import I make no questions asked.
There are different complexities, ambiguities, subtleties with other composition cataloguing systems that roon doesn’t handle well during the composition identification process. I find frequent errors with Haydn (Hob. numbers), Telemman (TWV numbers) and I forgot to mention Busoni (KiV numbers) where I had made a recent project to tidy up my library.
These issues are not limited to just these composers. With larger libraries these composition misidentifications may remain hidden for a long time unless you are double checking in the first instance as roon is not providing any disambiguation warnings during the importation process.
Thank you for taking the time to explain, tripleCrotchet.
I’m not really engaged in the various numbering schemes (seems like the Debussy one, in particular, was screwed up by Lesure, so that ANY database system would have trouble disambiguating it), just the names, keys, opus numbers, etc., of compositions. And I’ve not really seen much problem with those but, obviously, your knowledge is deeper than mine, and I really appreciate your taking the time to clarify
That’s very true, even if your albums are provided by streaming services, because of missing or erroneous metadata. Local libraries can be groomed, libraries provided by streaming services not so much. Of all metadata issues, the wrong or missing composition identification is the most vexing one…
Unfortunately, I have to agree. I find exactly the same types of composition identification errors with streamed Qobuz content but with no or much fewer options for manually fixing.
It seems to me that there are no real universally agreed and enforced meta-data standards so there can be no such thing as a zero-grooming experience except where users choose not to groom or as in this case roon imposes a zero-grooming experience. As an aspiration it sounds commendable, what’s not to like. But it does seem highly unrealistic. Fully onboarding the lack of meta-data standards would probably have led to a very different roon. As it is roon often feels like neither fish nor fowl. Too configurable for some and not configurable enough for others.
I only gave the Debussy example because he is a popular composer and the issue is easily understood. In most cases, also, the ambiguous L numbers will have come straight from the labels themselves or downloaded from popular download sites like Presto Music rather than introduced by users.
But TBH I play much more Baroque music than Debussy and there the issue is completely different with Telemann’s much more prolific catalogue. In this case the main issue is that the TVM numbers are case sensitive:
This causes havoc for roon and frankly the less familiar the material the higher the frequency of roon identification errors with all sorts of composers. Again by far the least line of resistance way of correcting this is to copy and past the allmusic structures in order to force roon to identify the correct Telemann composition.
I don’t have a solution to the unexpectedly high levels of grooming roon is observing. That is not my point. My point is that roon doesn’t have a solution either but carries on as if it does with an unrealistic design goal of no-grooming. I’m pretty sure most of the allmusic grooming roon observes is just to compensate for identification errors which are much higher than most people realise. That’s certainly my case, I’d rather not do it but I also want to be able to surface all performances in my library.
.
