Acoustid Albums Report and Import Feature

Hi, AcoustId currently has about 63M acoustids, MusicBrainz has about 25M recordings, so clearly there are many tracks in Acoustid that do not have a corresponding track in MusicBrainz

When Acoustids are submitted the basic metadata is stored with it, and when you lookup the Acoustid on the website you can see under the user submitted section columns for artist, title and album. What is not so apparent is that Acoustid also stores albumArtist, track no and year metadata as well, and track duration can be derived via the fingerprint.

As I have a combined MusicBrainz/Acoustid database I have been looking at this data and have been able to create Acoustid Albums that dont seem to be in MusicBrainz from the metadata.

The reports let you look at potentially missing albums, click on Show Tracks to see the tracks in the release and Import Release to seed the MusicBrainz release editor with the basic data, so I hope people find this useful.

I filter out poor quality metadata (e.g title = ‘Track 01’), and filter out albums where I could not find all tracks, I then filter out albums where more than three Acoustids are already linked to a Mb Recording because I’m trying to find albums that are not currently in MusicBrainz. Then for the albums left I do a case-insensitive search in MusicBrainz for albumArtist/Album to find matches that just dont have acoustid match and filter them out. Then I filter out any albums where I could not find a case insensitive text match between albumartist and an artist in MusicBrainz because this helps weed out some nonsense data.

Note:I haven’t tackled multi disc albums yet, each disc will probably be listed as a separate album, and some albums will be listed as multiple times, the compare with MusicBrainz is based on text match so in many cases the album may alreay be in Musicbrainz with slightly different name so proceed with caution.

The initial list is about 500,000 albums

I’ve split into three reports, based on how many existing releases the artist has in MusicBrainz. So the first one is artists with more than 100 releases so I expect most people will recognize the majority in the list, and alot of the missing releases are live albums, and also many classical releases. This report is more likely than the other two to list albums that are already in MusicBrainz but just with a slightly different name.

http://reports.albunack.net/acoustidalbums_report.html

The second is artists with between 2 and 100 releases, so these are smaller artists, and there are more missing original studio release but also plenty of live albums

http://reports.albunack.net/acoustidalbums_medium_report.html

The third is artists with one release or left, these are the really obscure artists, probably mainly self published releases.

http://reports.albunack.net/acoustidalbums_low_report.html

thanks Paul

7 Likes

Rebuilt with better matching against existing releases so that less likely to list albums already in the database, this is mostly a problem with the report for most popular artists.

Album Artist link takes you to the MusicBrainz Artist page for that artist, but only a link if that is the only artist with that case-insensitive name in MusicBrainz, because if there are multiple artists with same name we cant reliably identify the correct one. Similarly if you did Import Release it will use the Artist Mbid if can find unique name, but just enter name if cannot.

The Show Tracks button lets you see all the tracks, provides hyperlinks to each acoustid, and if any of the acoustids already have a link to a MusicBrainz recording the Has Mb Id takes you straight to that recording on MusicBrainz. But remember for any album the maximum number of Acoustids that can have a link to Musicbrainz is 3, the majority have no link to existing MusicBrainz recording id. This is to allow for the typical case that the odd song has already been matched to MusicBrainz because part of a compilation that has already been added to MusicBrainz but the original album has not.

1 Like

Made a few more changes:

  1. Removed albums that look like they are part of a multi disc release, as we don’t currently have a way to import a multi release it just ends up being clutter
  2. Added section at the top of each page showing what artist are on the page
  3. Added View in Albunack button, this is a discography page comprising combined MusicBrainz and Discogs releases for an artist and links and potential links between the two. If you are reviewing a particular artist it is often worth looking at this because the same release may also be in Discogs and this page allows you to import/seed the release from Discogs. Since Discogs usually has better metadata including coverart it is usually better to import from Discogs rather than direct from Acoustid
  4. If there are multiple artists with same name as albumartist instead of showing name as plain text we now provide Musicbrainz Artist Search link for that name, the link says Search ArtistName
1 Like