I think many are missing the point. I’ve posted about the absolute insanity that is non-existent metadata many times. The meta data “system” is a failure, e.g. there needs to be a standard which by artists, producers, labels follow and there “isn’t” that I am aware of. Add to that, “generally” speaking most/many people don’t give a crap, push play is all that matters, which is fine I guess
However, there are many that care about the entire production of an album, because that is what it is!! It’s not a single entity/person/singer. Far from it. I honestly don’t understand how anyone wouldn’t care about that aspect (all the meta data). But to each their own.
If you want proper meta data, you need to go get it, find it, research it - full stop.
Here is ones take on this specific issue > watch this Rick Beato vid, if you want the gist of what Björn states, go to 27:16.
I think that this thread has succeeded in drawing attention to the growing problem of missing or lack of meaningful metadata for many new releases and older releases from less popular artists.
The cynic in me wants to believe that the music providers (musicians, engineers, producers, labels, etc.) are none too pleased with the rapid shift to music streaming and therefore refuse to provide meaningful metadata to the music streaming providers. In other words, if the consumer wants metadata then buy a physical copy of the recording.
The realist in me knows that it’s more likely about greed and laziness. Greed because providing metadata (at the musician and label level) takes time and time is money. Laziness because seemingly everyone else is not providing metadata so why should they.
Basically between the lack of meaningful metadata for many recordings and the large number of unidentified (by Roon) COMMERCIALLY released recordings my Roon musical discovery experience is slowly grinding to a halt. Of course using Roon I can still “discover” the hell out of 60 year old jazz recordings.
I’d not use those derogatory terms. Most independent artists, producers, and microlabels barely make ends meet. Creating metadata gives them no value, and takes time that they could be using to make an actual living (such as teaching an instrument, which is how great very active jazz guitar players like Mikes Okazaki, Julian Lage, or Marc Ribot do).
There’s absolutely no simple answer to this issue. There’s so many levels of complexities involved here as to make this a Herculean effort. My answer is to do it myself. I learn so much information from multiple sources. It’s actually enjoyable. It’s a small hobby to me.
I recently put together a screen saver of more than 400 album covers dating back more than 100 years. For you jazz aficionados, I have a cover of a Blind Lemon Jefferson album. The screen saver plays on our 75" TV while listening to music. There’s a web site that scrapes the iTunes library for cover art. I haven’t put an artist in yet that multiple album covers haven’t come up.
I just run Bliss every now and again and it updates Metadata and missing images on my SSD if available it’s very reasonably priced. One very satisfied customer
The problem of metadata and their recovery is a long-standing problem and, I believe, without a solution. There doesn’t seem to be much interest from the music industry.
In the digital versions of the albums, except for some genres, you won’t even find the PDF booklet and the credits are reduced to the bare minimum. The physical versions, in more or less detail, the booklet should report this information.
For underground artists and bands, but often also for mainstream artists, it is impossible to find something acceptable if there is not some volunteer who uploads the information taken from the booklet or from other official sources, in the various Musicbrainz, Discogs…
But even these sites have limitations in terms of data structure and “certification” of the information, apart from the trust in the editor who wrote them.
Artificial intelligence can help make connections between entities or develop information, but if there are no detailed sources to draw on, this tool will not be particularly useful either.
I personally curate my collection using Musicbrainz. For each album I buy I go to check/insert the relationships taking them from the booklets, then using Picard I update and customize the files.
As far as I overview the situation, the standards set being Tivo and MusicBrainz are pretty precise and a lot of their data is actually good. Otherwise roon would not work so well with common releases.
I see a problem in many small record labels and independent publishers not taking this seriously. And some majors are sitting on an ocean of data but they do not care a lot about its consistency so just upload it unchecked.
What I find annoying is especially streaming services basing their offer and search routines on inconsistent and contradictive metadata (Tidal in particular). My understanding is they have accepting all content in the past to simply offer a lot and now there is a desperate need for stricter guidelines.
Maybe one day Tidal, Qobuz and roon could offer some routines that allow users to simply hide albums without proper metadata. That might function as an encouragement for labels to complete their data otherwise they loose streams and revenue. I have the feeling that Spotify and Apple Music are trying something similar already as their base seems to be better.
That is likely to be the actual perception, yes. But if they understand that they are found my more users and their streaming revenue is instantly increasing, they might be encouraged to provide better metadata. And I am sure they could see it as in investment into fan loyalty as well.
Is the majority of your albums not identified due to inconsistent metadata leading to them not being recognized automatically by roon (but can be identified manually)? Or are these albums in the majority simply not existing at all on MusicBrainz?
In my library I found the number of albums being manually identifiable surprisingly high. But I do have completely different genres so situation might be different.
Not sure they do, I am finding inconsistencies regularly where they mix two or more unrelated artists togther. Flag this to Joel who then has to then ask them to fix it. This happens a lot more than you think.
Adding new releases to MusicBrainz is relatively painless using Picard as a starting point if they dont exist, but adding new artists, contributors etc takes a lot of time and new additions tend to go unchecked so mistakes can happen. Editing existing releases or release groups, artists etc there is more of a system in place to validate whats being changed, and votes are required for edits from the main people monitoring it all. This does lead to some narrowminded views at times and you can quickly get frusrated by this process. Had a few recently and just gave up trying to change them for the better. But overall this process works reasonably well.
The metadata doesn’t need to be created since the information that makes up the metadata is already exists in the form of the recording session logs. Really how time consuming would it be to include the names of the musicians as well the engineers and producers and the date and location of the recording. Here’s a fun fact - the original specification for the compact disc included the ability to have the track titles, as text, embedded on the CD and some early CD players were able to display the track titles when playing the CD.
I don’t think that I’m being all that demanding asking for the names of the four musicians who played on the John Doe Quartet instead of just “The John Doe Quartet”.
The approximately 10% of my music library which is unidentified by Roon is split between unofficial (aka bootleg) recordings, which I don’t expect Roon to identify, and, as I stated earlier, commercially released recordings, which I do except Roon to identify. For many of the commercially released I often have the names of the musicians listed in the file tags. The unidentified recordings situation is similar to the metadata issue since it’s another blame game - Roon takes my money and then fails to deliver, blaming MusicBrainz. Not cool.
No, your chronology is off. That is/was CD-Text, an extension to the Red Book standard. It was not added until the 1990s, about 15 years into the CD era.
Are you expecting Roon to move into the metadata business and employ a large team of researchers / cataloguers (across a wide range of musical genres) to research and then add metadata for any albums in a user’s library that aren’t matched by Tivo / MusicBrainz? That’s putting aside any licencing issues the potential sources of additional data might infringe on, see Band Camp example above.
Doubt that will ever happen with Roon, or any other music service that charges $12.50 a month for that matter. It simply not enough to cover the costs that task entails.
Sure, the record companies could build better processes and pipelines to extract and feed more of their existing data to the various metadata providers, but that’s hardly something Roon have control over.
Where Roon could maybe help is in leveraging its existing userbase to crowd source additions or edits, much like it does with images. Potentially even feeding that back into a service like MusicBrainz where it can be further linked up and potentially improved further.
In lieu of that, realistically the best soluton is probably to download Picard and submit the changes to MusicBrainz yourself. Not only will that improve your library, but if I and others have some of those albums we benefit too. Likewise, if I and others make some edit to MusicBrainz and you have those albums in your library you benefit. Far more productive that shaking your fist at clouds
So here’s where I do think there’s the possibility to make this into a business model, albeit one that would be marginal in financial impact at best (because it would create an asset that by definition isn’t that valuable).
Provide simple ways for Roon users to edit metadata in a more “real-time” fashion - eg, when looking at an album, enter a metadata entry mode where you can click on and edit fields, add composers, etc. Individual users can choose whether to share all of their edits, or whether to selectively share. You can gamify contributions if you want, or choose to provide some small value tokens. Aggregate and have voting visible from the Roon interface (likely somewhat similar to what Valence did with images, obviously a bit more extensive and complicated to manage the interface). Get reputational effects for users with good contributions. Arm the editor with AI-generated suggestions for users to approve or not. While the challenge of “generate all the metadata for this album” is a very hard problem, finding potential improvements is a much simpler problem, and to the original point that @Jazzfan_NJ made, filling in an absolutely empty set of disc information in a way that you could get track names and some artist credits wouldn’t be 100%, but if it was “AI suggesting to human within an interface that they’re already using to make edits themselves, so it’s just like another suggestion”, that’s not that hard - and very similar to what others are doing. Yes, you can get model collapse and junk if humans are ok’ing junk data, but that’s where reputational effects come in.
Then you’re left with whether to contribute all that data to Musicbrainz, or to create yet another Roon protprietary repository. Obviously some people might not contribute based on the principles that @Jazzfan_NJ has - don’t do free work for for-profit companies. But many many would (witness Valence and the very impressive impact it’s made) because they’d contribute some and gain a lot.
Now, would I do this if I were in charge of Roon? No way. Way too complex, and way too many things that would bring in revenue. Most people (by definition) have mostly popular recordings. Even I, despite my desire to think I’m really quirky in my taste, have vast majority well-metadata’ed recordings. So I don’t think this is worth it. But I do think it’s a good idea.
You points are quite valid and well taken. What I would like to see Roon do is to exert some pressure on the streaming services to provide at least some useful metadata with new releases. As I keep repeating the lack of metadata makes for a very unfulfilling Roon user experience. Sure going back and trying to fill in the missing metadata for so many recordings is a monumental task and I don’t expect it to every happen but, going forward, providing metadata for all, or at least most, new releases really isn’t that arduous at task, all it requires is a little effort and a lot of will power.
Let’s try thinking about it this way. Music streaming is now the main way that new music is distributed to the listeners, replacing downloads which in turn replaced CDs and LPs. If one purchased a CD and there was no information in the package other than the song titles and the name of the group, e.g. The John Doe Quartet, one would not be very happy. Why are we all so willing to accept this lack of information when it comes to music streaming? I’m not asking the music streaming services to provide us with full, detailed recording and discography data but just something similar to what was provided with the physical medium.
At least Tidal doesn’t seem to be a part of this at all. Not sure if Roon has any leverage to pressure. And Tidal may have contracts with the labels that prevent it, so they may as well not be in a position to change that.
So the upshot is that Roon needs to revise its marketing information to reflect that while one can link recordings from Tidal and Qobuz into one’s music library, those linked recordings may not have any meaningful metadata and therefore will not provide the full Roon experience to the user.
And yes, I am beating a dead horse but the horse is dead and can’t feel anything so I will just continue beating it because it makes me feel better.
Hey Nick, Grateful to you for posting this idea. For <$100 for unlimited, this may be my path.
I bought an evergreen Roon license in 2021 for exactly the marketing promise of metadata and what I thought would be a constant update as new Album information became available. My significant other thought it best that I not maintain the jewel cases of precious covers, liner notes, etc. Roon to the rescue. I have not been blissful since.
