Track editing, auto editing tracks and editing all tracks with identical title (Last.fm features)

ListenBrainz
1

Last.fm recently fixed one of their long-standing issues, by implementing auto editing tracks:

This is especially relevant on Last.fm, because previously, if you scrobbled through a music player that lacks support for for scrobbling album artist (Spotify being a good example), and the track artist is something like Michael Jackson featuring Janet Jackson, the album artist field for the same track on Last.fm got the same field as the track artist - Michael Jackson featuring Janet Jackson, when it should be simply Michael Jackson as the album artist. So now you can auto edit rules that whenever you scrobble a song with Michael Jackson featuring Janet Jackson in the album artist field, Last.fm will automatically change it to Michael Jackson, for that specific song. Or it will change Michael Jackson featuring Janet Jackson in the track artist field to Michael Jackson, Janet Jackson (which is what it looks like when you use the split artist feature in foobar2000, and some other players).

I think it would be very useful if ListenBrainz also implemented the auto edit feature. This is especially relevant because while this is no longer an issue on Last.fm, it’s still an issue here, and while it seems like ListenBrainz doesn’t make use of the album artist field, it’s relevant with other music players like AIMP, where the split artist feature lists multiple artists as Michael Jackson/Janet Jackson, which as far as I’m concerned is bad optics.

As for track editing and editing all previous tracks of the same song with identical titles, it would also be great if ListenBrainz could implement this here as well. Why this would be useful here, is because when we import our scrobbling history from Last.fm, it’s often the wrong casing and such, because Last.fm doesn’t recognize how the titles are written from your scrobbles, but rather, from the first person who scrobbled the song. So if the first person who scrobbled take say, 2Pac, Snoop Doggy Dogg - 2 of Amerikaz Most Wanted, and he did it as 2pac, Snoop doggy Dogg - 2 of amerikaz most Wanted, that’s exactly how it will be listed in your Last.fm scrobble history as well (apparently it’s a server issue on their end, might be fixed later), and we can’t edit the casing on Last.fm for tracks, just change values such as album artist and so on, and so when we import our Last.fm history to ListenBrainz, we get the same casing errors here. But not only that, Last.fm also has different MusicBrainzIDs for various songs and albums (lots of scrobbler components/apps don’t even recognize these fields when scrobbling, and I’m not sure if Last.fm even follows MBIDs; if the artist, title, album and album artist fields are identical, the track in question is scrobbled to that page regardless of whatever MBID is listed on that page). If I remember correctly, Last.fm does have MBID, but it’s not scrobbled to Last.fm, and whatever MBID is on the Last.fm page, that’s what you get when you scrobble on Last.fm.

Anyway, would be great if ListenBrainz could implement these editing features. That way we the users, would have more control in tidying up our scrobbling history with more accuracy.

1 Like
2

I think letting users edit their listens is probably a good idea, however there are multiple issues with it that tie in with LB implementation details [1] so I wouldn’t expect this to be done soon (I may be wrong though will need to discuss with other devs to confirm).

However, LB does not suffer from many of the issues you mentioned about LFM data. In LB, we store the exact same data [2] as submitted by the user/player/service. On the listens page, the data as submitted by the user is shown. Everywhere else, the feedback page, the pinned recordings, statistics so on use data from MB if possible otherwise fallback to user submitted data [3].

To use data from MB, it is necessary for the listen to have MBIDs submitted with it (it is definitely preferred though). LB runs a mapping process and which tries to find a match for the listen in MB and links MBIDs to it. The process is case insensitive and works even if there are some typos in the submitted listen title/artist. We consider the data in MB to be the correct data so it does not suffer from casing, typos issues alleviating the need for editing listens.

[1]: Letting users edit title and artist of the listen, especially has more issues than other fields. For other fields, its easier but a matter of implementing it to scale well.

[2]: We remove some fields like mbids from the submitted json if they are empty but other than that the submitted data is not modified.

[3]: The current state of which page uses data from which source is a bit complicated but this is the goal and should soon be true.