NickvisionApps / Tagger

Tag your music
https://flathub.org/apps/details/org.nickvision.tagger
MIT License
231 stars 14 forks source link

musicbrainz integration is very imprecise #48

Closed legacychimera247 closed 2 years ago

legacychimera247 commented 2 years ago

i love the fact we can search for tags in musicbrainz (really love it) but i need to say that more than 50% of the times it picks the wrong record, even if i choose all songs from an album (that maybe are parts of various compilations ond so on, the app will probably choose the compilation over the original album, that maybe is already in tags. Oh, and it's also seriously slow

Anyway to correct this? :)

nlogozzo commented 2 years ago

The way the Download MusicBrainz Metadata function works is Tagger uses the chromaprint fingerprint for the file and sends a query to AcoustId, which returns a list of MusicBrainz records, ranked from 0-100 (closer to 100, closer the match). Tagger then picks the highest ranked record and sends a query to MusicBrainz to get all the information it needs for the tag. On top of that, Tagger will also query CoverArtArchive to get an album art for the album provided by MusicBrainz and will download that image and set it as the album art. I will try to compare our MusicBrainz metdata action to that of MusicBrainz Picard and see how Picard picks a record and try to improve the accuracy of picking a record

nlogozzo commented 2 years ago

As far as being slow, from the above description you can see that each file makes about 5 internet requests (AcoustID Query, MusicBrainz Record Query, MusicBrainz Release (Album) Query, CoverArtArchive Query, Downloading Cover Art Image) This starts to pile up fast, for example, running the 'Download MusicBrainz Metadata' action of 5 files is about 25 internet requests made, which depending on your connection can be slow. The best way that I can think of now to improve the speed is to maybe run up to 3 files on different threads handing about 15 requests at a time instead of 1 at a time per file.

nlogozzo commented 2 years ago

Speed and accuracy should be improved in V2022.10.1-next1 Please test and let me know how it goes.

Side Note: Download MusicBrainz Metadata action has been moved to a new Web Services button next to the Tag Actions button Screenshot from 2022-10-02 13-31-03

legacychimera247 commented 2 years ago

hello, this was much faster i need to say, only problem is, since not all files as the acustid, some files are missing

as for the speed itself it was much faster, need to do further tests anyway

nlogozzo commented 2 years ago

To fix the issue of a file not being found with AcoustId, i added a feature in V2022.10.1-beta1 called 'Submit to AcoustId'. This will allow you to select a file and associate it with a MusicBrainz Recording Id that you can get from online.