I’ve just posted an article about importing music libraries to Roon:
It’s basically a bringing-together of all the main references in one place, and the extrapolation of some “maxims” to guide you when organising your music library.
The way I wrote this was basically to import one my own libraries (admittedly: a fairly messy one I use for testing) into Roon to see how it fared. After doing this, and seeing the problems, I came to the conclusion that organising your library is still important to maximise your value from Roon.
I wondered if anyone had any thoughts on the article - anything obvious I’ve missed?
Disclaimer: this is on the website of my commercial software product.
Good work that addresses many important points of view.
Since Roon is primarily based on AllMusic and MetaBrainz, it falls short of the really great databases. I have certainly implemented all points of view with Foobar2000 and still the 80:20 rule applies (80% without identification). I have about 80,000 of the many millions of artists in my library and see text and image gaps not expected even in the mainstream. It gets better and better in the timeline, of course.
The world has changed. After a long time of operation, Magix stopped the operation of the free CD database freedb. It all started in 1998 when the company CDDB (Compact Disc Database) was founded. After a sale to the long-defunct Hi-Fi brand Escient and renaming to “Gracenote”, freeDB became more popular. Gracenote changed the license terms and whenever everyone thought only about the money it crunched in the organization. EAC or MP3Tag don’t access it anymore. Musicbrainz, AllMusic, LastFM, Discogs or Amazon were the new hopefuls. This order almost gives the death knell, whereas Amazon still seems to me to be the most alive. Even Roon hosts there (but probably with less joy for the customers and without metadata service).
The new home for the freedb.org database is now the free GNUDB accessed by many rippers and taggers. Discogs is also very broad with its LP history and has become the reference model for many that the Musicbrainz project is just too academic, perfectionist and bureaucratic on the spot.
Of course, there are very many specialists to explore. Soundcloud for the masses, odesli.co for simultaneous searches in several services, the AlbumArtDownloader that rummages through many dozens of providers and thus simultaneously reveals most of the special sites such as 45cat. In the end, you end up at Memoryradio, where historical German songs are in the best care and collection.
Unfortunately, the best source of metadata and information remains the booklet. As mentioned in other discussions, streaming services, record labels or commercial databases do not have, and probably never will, the detailed credit information that you will find in the booklets. There is no interest in this (just look at the digital booklet: especially for certain genres it is very rare to find it in digital releases even if ALL these are created as digital files and then sent to print for the various physical releases of the disc. There would be no costs by attaching it to the digital release, but it doesn’t happen).
Consequently, the only way to get a good detail of this information is to rely on the desire of users to enter what they find of the booklets and the availability of opensource databases. The two biggest are discogs and musicbrainz. The first has the advantage of having many more releases considering that it is also a buying and selling market. The disadvantage compared to musicbrainz is that the insertion of credits is much more “free” having fewer constraints on how to write this information (and this is not good especially for those external programs that go to retrieve the information from their API) and there are fewer checks on the correctness of the information
This absolutely depends on the booklets. I have many albums where Discogs as well as Roon knows much more than the booklet. Some bands have fan sites that collected much more info than the label provided in the booklet, which typically makes its way into Discogs and others as well
You say that, but there must be some interest in this (you’re mentioning it for starters). That means it’s down to the economics of whether it’s worth delivering on that interest, given the size of the market.
Obviously. Especially for famous albums there are “encyclopedias” of information on the internet.
But most of the time, if you want to know in which recording studio a song was recorded, or who played the flute or violin on a certain track (and who is not in the band), the booklet is the only one source where to find this information
Thanks for the mention of xLink - not seen that before.
I should mention I do run my own metadata service - http://www.onemusicapi.com/ - so I’m not unfamiliar with what you mention, I was just wondering if there were particular providers you saw as being better than others.
What I am saying is that it depends, so the statement “Unfortunately, the best source of metadata and information remains the booklet. As mentioned in other discussions, streaming services, record labels or commercial databases do not have, and probably never will, the detailed credit information that you will find in the booklets” is not universally true.
Sure. But for what I see, and for the genres I listen to, that’s it. And I clarify that I speak of “official” sources, so I leave out everything done by the fans, by the artists themselves or by “experts in the sector”. These are sources that are too variable and depend on the individual case. They should be in addition to the normal credits.
If we want to make a parallel, let’s take the movies: we all expect to find the credits at the end of each movie (and in 99.9% of cases it is). If we want to know the name of the photography director, we know that we can almost certainly read it there.
I don’t disagree that it would be nice if labels had published complete official credits, and had always done so since the beginning of recordings. But it would be too much to expect from, say, a small band and indie label in the often heady times of their careers. The film analogy is that you get full credits for movies made through established channels that make it into film theaters. You don’t necessarily get them for every indie movie, and not even for, e.g., Netflix productions
With 8000000 albums and 4000000 artists Roon would be well advised to take a look at this service, because the average provider has with 80 to 90 million titles (many multiple counts) also not more albums or artists, rather less, because much of what is only released on LP has no streaming rights. The allocation problem we have yes even still as a rare species with CD releases.
I have just been trying to make a copy of my music library for the car and came accross this thread. My library started off in iTunes, go converted to a Roon core on Mac, then a NUC, then a Nucleus. Ripping CDs to Roon has been a trial as the CD ripper produces a CDripped with date entry in the Library but with no information about the content at all. I did find the CD ripper in Roon almost useless. Other albums have been ripped via iTiunes which still works OK. But the history of all this gives me a library with miscoded files, some weird folder structures and all kinds of chaos. Its a tribute to the power of Roon that I was blissfully unaware of this chaos under the surface but having found it I really need to sort it out. First priority is to to label these CD rips, then to get the rest into a better state. I looked at MusicBrainz and mp3tag to label the unlabelled rips but cant get them to work. Bliss looked promising but the interface is a bit unintuitive. My plan is to produce a cleaned up library on external hard drive and then use that rather than the internal drive in the Nucleus. It is compact and space saving but does complicate backup copying as it had to go on over the network. The last Library backup took 52 hours to make a full copy of around 2Tb . Anyone have any suggestions?
If you export the albums you ripped in Roon it will add the metadata. Although I don’t think you can do more than a couple at a time.
The main advice from fellow users has always been rip your CDs on a computer using a good ripping software, I use dbPowramp, and then after double checking the rip, copy it to the Roonserver.
My experience with Roon fluctuates between frustration and fascination.
I think your metadata is already fine, now just comes the step of getting the already enhanced to an external hard drive.
Have you tried to solve it with Roonexport? There are three possibilities to export the files themselves, only the metadata as CSV (for Soundiiz) or the metadata to Excel (spreadsheet).
I would never have thought of copying the data itself over the network with Roon on a weekend.
The times of greatest user load then do not allow the good user experience of the other times. Are there better values at other times or did it remain with this one disappointing attempt (with 52 hours for 2 TB)?
Do I imagine correctly that the data has been copied from an NTFS disk internal (Nucleus) to another NTFS disk external?
Were other formatting (for Apple / Linux) in play?
Would not it also be possible via USB? External hard drives can be connected to the Nucleus. Is this only USB2.0? I do not have a Nucleus here.
This is a great article, thanks for sharing the info… For me, I still think it’s a real pity that Roon does not make use of AcoustID’s in order to better identify the music in the library, even when it’s not tagged perfectly. Many albums which are recognized by Musicbrainz Picard without any problems don’t get identified in Roon without manual intervention. When I first started with Roon I had around 9000 unidentified albums in my local library. After many months of manual work submitting literally thousands of albums to Musicbrainz in order to help Roon identify them, I am now down to 6000, so this seems like endless work to me. The main reason I was attracted by Roon to begin with was the promise of vastly improving my library browsing experience by collecting new metadata from external sources. It works well for mainstream albums and the result is impressive indeed, but for about one third of my library, it failed to provide any useful info. Needless to say, that’s the third for which I am most interested in getting extra information as it mainly consists of obscure artists about which I would like to learn more. Many of these releases come from Bandcamp and yes, I can import the metadata from Bandcamp to Musicbrainz and wait for it to become available in Roon… but at that point the metadata is already in my files after tagging them with Picard, and Roon will not find any extra info on top of what I upload to MB, so why do I still need it?
So at this point I am still not convinced that Roon is the way to go for me, after trying it on and off for a few years already… I am currently sticking to my LMS installation which I’ve been using happily since 2008, after tagging my files with Picard. I love the Roon interface, but as long as it cannot provide improved metadata for at least 95% percent of my library, I think it fails to deliver on its promise.
