Before I address your points directly, some of the higher-level view…
These are areas that we’ve worked on intensively over the past two years, generally in two areas:
- By making export functionality more complete
- By expanding support for reading metadata from file tags, and controlling how that data is used.
These go hand-in-hand, but for a serious metadata groomer, the second is the more important of the two. If you are interested in durability/portability, lean towards fixing how things show up in Roon by working on the directory structure or file tags using external editors. With respect to many of your concerns it is possible to solve problems this way and keep those changes outside of Roon’s database.
It’s possible to control how tracks are clumped into albums via directory structure. It’s also possible to tweak file tags on disk in order to motivate certain identification results, multi-part work grouping, and many other aspects. In some cases it’s necessary to use “prefer file data” features in Roon in order to get your data to “show through”. By approaching it this way, you create durability implicitly.
From our perspective, file tags are not a good representation for music metadata. They have limited expressive power, inconsistent capabilities from one file format to the next, and wildly varying common usage. It’s possible to create a certain familiar, lowest-common-denominator experience from file tags, but that’s about it. So for us, tags will continue to be there for interoperability–something that we handle at import/export boundaries, not the main place where the data lives.
We built Roon in large part because we are believers in automated metadata grooming. In some ways, thinking about how work done in Roon will “port” back to software built around the old paradigm is kind of like asking how to play CD’s on a phonograph. Having lived with automatic grooming systems for as long as I have, I can’t imagine going back to the old way. I assume that anything that would compel me to switch from Roon would do this stuff even better than we do, and that I would care about the time I have invested into file tags even less than I do now. :). In other words, our mission is not to trap your effort, it’s to minimize it.
Our database format is closed because we require much more expressive power than file tags can provide, and because we need the freedom to evolve the schema and database implementation over time. If there were a third party ecosystem reading/writing/understanding that schema, every decision we made would be set in stone from the moment some 3rd party depended on it. It would not be a healthy environment for growing a product.
Onto your core points:
To the point: I believe that any collection grooming work needs to be exportable with the media.
I agree with you–to the extent that well-established conventions exist, and meaningful portability to other software results.
There are limited options for this, and the obvious one to me is creating custom metatag fields that can hold multiple values - “Tags” is the obvious one,
Exporting “Tags” is a no-brainer, I agree. We have a work item in the queue to go through and support export of as much data as possible. This will be part of it.
but there is actually quite a substantial amount of media information that the user has to correct/input to get Roon working perfectly, and the vast majority could easily be stored in custom metadata embedded in the media files.
As far as I am aware, ideas like choosing preferred versions + identifying albums (the other two examples you mentioned) do not map in a straightforward way onto existing tagging conventions or other software–what are you thinking we would do here?
There are some straightforward things we could export (like LABEL
and CATALOGNO
) which would allow us to express the idea of an identification in a way that was not so tied to identifiers private to our world.