I notice that when I sort my albums by “most played”, that larger box-sets are over represented. For example, a box-set with 10 CDs of 100 tracks that I have only played through once is considered much more played than an album of 10 tracks I have played through 9 times. I would give much more weight to the single album.
Does this bias roon recommendations for radio, recommendations for you etc.? In my case I don’t think box sets are particularly representative of my listening tastes. The only reason I can think of why I would have played one through is that I left roon playing overnight or otherwise unattended. In any case, box-sets also have quite misleading genre metadata as it is album rather than track specific. I just wonder is this effecting roon automation such as radio and recommendations in any way?
Generally, “Most Played” should be a representation of what you have spent the most time listening to rather than the total number of album plays. So, for example, if you have a large box set that you listen to once, and a smaller album you listen to twice, chances are the box set will appear first when sorted under most played because more time was spent listening to that box set even though you only listened to it one time through.
I was afraid of that. I am not concerned about the way that “most played” is sorted on the album page. It is more is this information used by roon in radio and/or “recommendations for you”?
The problem is I cannot see that “most played” calculated like that would be a great way of calculating “most listened” to. Even a modest sized box takes all day to listen to and larger ones might take a week or two. So I know I don’t do that. Does anyone? The only reason I can think of why several would rise to the top of my most played was because they were in background or running overnight and I haven’t switched roon off or changed the disk. So I am not really “listening” to them. A conscious choice to play a shorter disk all the way through 10 or more times is a much more important reflection of what I am “listening” to it seems to me.
Two other things. First, my experience is box sets tend to have very generic album level rather than track level genres that often also don’t really reflect conscious listening habits either. For example with Classical I am often surprised that radio and “recommended for you” pick so much large scale symphony and orchestral music I have very little interest in listening to in a home environment rather than the smaller scale chamber, piano and vocal music that I am. It just seems as if all “Classical” is put into some very coarsely grained bucket of “Classical” like it is all the same.
Second my experience of radio and “recommendations” is that I am surprised by the bias towards Classical. I do listen to it, but far from exclusively and I just wonder if the Classical (and Jazz) box sets that I am not really listening to but nevertheless are rising to the top of my “most played” are skewing roons picks in some way?
The way they have calculated “most played” pre-dates the Roon Radio recommendations.
The Issue is not so much with Box Sets, but longer works vs. shorter ones. For example, listening to an Wagner Opera instantly gives that listen potentially 4 hours even though you’ve only listened to it once. While the 3 minute track you’ve listened to 50 times is still in a distant 2nd.
Interesting. Probably time to revisit the algorithms surrounding this. I’m finding other oddities. I can see otherwise lightly played albums rising to the top of “most played” because I might have a lot of different masters. It is almost as if a played once album is multiplied by the number of masters in your library. Surely not? Again if having a lot of versions effects the calculation of “most listened” that is very miss-leading as well. In most cases where I have several versions it is only because they are available on Qobuz and I want to experiment and then never delete the duplicates.
I just wonder if all these factors are leading to picks from roon in various parts of the UI that don’t seem to make much sense.
I think that when you have multiple versions of an album and Show Duplicates set to off, the total play time for all versions of the album is used. You might test that out, Tony.