Using Roon for the first time - very slow to filter by tags

I am trying to set up Roon for the first time. I have around 50,000 tracks. I have disabled audio analysis for now. I have created four different tags. Tags which have large numbers of tracks are taking long to load, as follows:

Rating - 0: 19,000 tracks - 65 seconds
Rating - 3: 22,250 tracks - 80 seconds
Rating - 4: 6,450 tracks - 15 seconds
Rating - 5: 2,900 tracks - 5 seconds

In some instances, when I clear the tag, so that Roon should show all tracks, it is taking just as long to reload all tracks.

I have around 10,000 favourites and these load and clear instantly.

I am using a 2017 MacBook Pro which ain’t no slouch and Roon 1.7 (Build 521).

I am hoping that this isn’t the expected behaviour and that it is either related to Roon still doing background work as I am setting it up for the first time, or there is some other issue that needs addressing in order to get the expected behaviour.

This is a known issue. Roon can be really slow with complex or large tag queries.

1 Like

Oh dear. It’s actually worse than I thought.

I set up a tag query to exclude all of the 4 tags I listed above, i.e. I wanted to just see tracks in my collection that had none of these tags attached, and the query took 13 minutes! I wanted to set up a bookmark for this query, as it’s a query that I will often need to use, but each time I click on the bookmark I would need to wait 13 minutes - this is unworkable!

As someone who is thinking of moving from iTunes, where smart playlists load and update instantly, this is, unfortunately, a deal breaker.

I am aware that, once a query has loaded I could add tracks to a playlist, but the whole point is that I need a more dynamic solution - such that when I add new tags to tracks, the bookmarked view will update automatically. Static playlists won’t work in this instance.

Until they decide to make roontags a first class citizen it’s pretty much a waste of time using them for anything but arbitrary stuff. Performance was raised as an issue shortly after release, but seeing as its nothing to do with the music you don’t have or anything that can be leveraged in their machine learning code good luck ever seeing any improvements.

I have found a lot of use in Tags, but yes, due to performance, there are limits. A few observations:

—Direct Tags work best. In other words, tagging a group of Artists to shuffle Artists, or a group of Albums to shuffle tracks from those albums works pretty well. Those Tag queries/filter (whatever you want to call it) tend to be near instantaneous unless the Roon core needs a reboot (every 3-4 days).

—Conversely, indirect Tag filters/queries (finding an object by whether a related object has been Tagged - i.e. Tag the album and try to filter for tracks) can be painfully slow. For example, Tag 25 Artists. Then click “View All Albums.” Unless each artist has just 1-2 albums, this can take minutes to complete. It gets worse the larger the group.

—Tags are better for subdividing your collection rather than aggregating it. Meaning, breaking it down into smaller pieces works better than trying to Tag half a large collection for some reason.

—You can Tag Tags. This allows for a modicum of level-organization (since Roon is folder-phobic) and stacking Tags at levels may help you create larger groups, as long as you don’t try to query indirectly.

—Clearing Tags from a filter can take as long as the original filter application. This particularly is a head-scratcher since Roon is pretty snappy at displaying the library generally.

—Combining Tags, positive or negative, can be super, super slow.

The moral of the story is that Tags can still be useful.

I’ll rephrase… The moral of the story is that Tags have the potential to be useful.

Thanks @James_I and @evand.

My experience pretty much corresponds with yours @James_I. Direct tags are best, though still load slowly with large numbers of tagged files.

For instance a single direct tag for 30K songs takes 2 mins 20 secs to load for me, whilst a direct tag for 4K songs takes around 8 seconds.
And a single inverse tag for 20K songs takes 4 mins 20 secs to load.

Whilst sometimes it takes as long to clear a tag as it does to load it, sometimes Roon updates instantly when I clear a tag (even for 30K songs this happened on one occasion).

The vast majority of my tags will be for less than 5K songs, so manageable in terms of the time needed to load them, unless, as you say, I am combining tags, as this will slow things down again. But I will need to use a few tags with larger numbers of files (as I need live updating and playlists in Roon won’t do this). So I am really hoping that there is a way of speeding this up.

It seems that there isn’t a direct correlation between the number of tags and the time needed to load them - i.e. there is an exponential increase in the time needed. If 4K tracks takes 8 seconds, then 30K tracks should take 60 seconds to load, but in fact it takes more than double this time.

I look forward to hearing from the Roon @support team about this. I can’t get my head around why smart playlists update instantly in iTunes, but Roon tags are so slow. Still haven’t decided whether this is really workable for me.

Nick

Roon can certainly get better with Tags. Have you tried making a Bookmark from a Tag, and seeing if the process is faster by calling on the 0 Star bookmark rather than going to Tags and selecting the 0 star Tag?

No, using a bookmark doesn’t speed things up, unfortunatety.

Hi @nickharambee,

Thanks for reaching out to us here. We are aware of a performance limitations when using tags that are applied to a large number of tracks.

We have a ticket regarding this behavior and we are looking at improving the performance of these queries, but at the present time I can’t comment on when exactly it will reach the dev queue.

I understand that this is an area that is important to you, and I wish I had better news here but this is currently expected behavior when performing complex tag queries.

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.