Is this content from local files, TIDAL, or Qobuz?
Screenshot of import settings
Description of the issue
It would be helpful to have a pinned post from Roon explaining what one can expect when reporting metadata issues here. It’s not clear if reporting them does anything to correct them.
Is this topic just a place to vent?
Are errors reported to metadata providers for correction? If so, is there any follow-up by Roon? Is the user ever informed?
Does Roon correct any of these errors themselves?
Is the volume of posts here useful in pressuring metadata providers to do a better job?
Everyone has their likes and dislikes, but to me, the poor metadata on some releases is a huge frustration. It may be a small fraction of releases, but when it’s the one you want to listen to, that doesn’t matter.
I’m sorry that you are frustrated with the metadata on some releases.
In answer to your questions:
No, if a fix is within our capabilities, we’ll do what we can, though we often can’t give any timescales.
That depends very much on the nature of the “error”, but we do report issues to TiVo and the three streaming services that we currently integrate. Sometimes though, our requests might conflict with TiVo editorial policies, and streaming services often have to go back to the labels themselves to get things fixed (which sometimes never happens).
We do try to keep tabs on the issues and update the original threads when something is fixed.
In generally no, as we are not in the business of correcting metadata. Occasionally though we make an edit to the MusicBrainz database if that is the quickest and most straightforward way to fix a problem.
I don’t think so.
Regarding your specific issue – the Jerusalem Quartet album – we’re falling foul of poor quality metadata from Qobuz on that release. Specifically:
They alone have used a French title which clearly doesn’t match the album cover.
There is no composition information and the track titles just display the movement names.
Critically, someone made a transcription error in the track recording code for track 5. Our code has – correctly – picked up this difference and refused to join this release with other releases of the album which have better quality metadata.
I will reach out to Qobuz – in this case – and request that they fix the metadata.
Oh, you were referring to the manual identification wizard; that wasn’t clear…
There are indeed two “editions” available via the wizard: one with the poor metadata from Qobuz – which will be used by default because it is a precise match to the Qobuz release – and one which actually reflects the metadata we get from TiVo.
“Editions” in the wizard is actually a really bad name – it should probably say "Releases. (I’ve made a note to raise this internally.)
We have two different releases, each consisting of “equivalent editions from different metadata providers” here, because – as I explained above – errors in the Qobuz metadata for their edition of this release prevented it being joined up to the others. With correct metadata, there should only be one release. Apologies for the terminology, which is confusing.
@Mike-48 Qobuz have said that they are going to request a re-delivery of this “product” from their supplier and that that should fix things. We’ll see.
You say there should be only one release. My experience is the opposite with Classical anyway. There are usually multiple options and the default one is invariably wrong. Typically for the same reasons, composition identification and composition grouping.
My habit is to manually re-identify everything for this reason as accepting the default match leads to many problems.
But concretely, my question is, in this particular case should I be concerned with the metadata of the second edition/release in the manual identification wizard?
@joel What is the sort order for the different editions, or actually what determines the default? Release date (newest first), popularity, best match, the most distinctive edition (most types of entries and therefor with the highest chance of errors), something else?
I would say that with Classical it is sadly rare that I could use the release date as the sole decision criteria about which release to accept.
It’s great when that does work but I usually decide based on what I judge will result in the smallest amount of post-edits. Usually that means accepting whichever release identifies the most compositions and gets the composition grouping right. I also much prefer multi-CD releases compared with unnavigable digital downloads which is what I may well have purchased. Almost everything else can be post-edited afterwards more easily. For example, recording dates, release dates, recording location, label, catalogue number etc. etc.
So personally my sort preference for different editions/releases would be some measure of the likely post-editing involved that took into account many more data points than the current simple measure of track miss-matches.
I’d have to check the code, but I suspect that there isn’t a specific sort order. I think we should probably implement one based on the number of discs and tracks, and ensure that the “trackless” releases are at the end.
