flakie 11 Posted July 26, 2019 Author Posted July 26, 2019 (edited) Thanks, I will look at the musicbrainz Picard. I am assuming this just adds a tag that I can do myself manually/scripted in mp3tag (which is the software I use for all tagging)? I have now sorted all the duplicate artists and cannot remember exactly which ones, with which albums, were the problem. I think for Madonna it was as entered below. I keep minimal metadata in the files and is exactly the same format for each album (it can be seen in the below for Madonna). The only exception is if it is multiple discs I also add an entry for DiskNumber, in the format 1/2 etc (otherwise DiskNumber is blank). I always enter Artist as the artist who actually performs the track, which might be different from Album Artist and may be many different artists on each album or each track. If multiple artists for each track they are separated by a semi-colon. I always enter AlbumArtist as whoever is the main artist on the album. This is always exactly the same for each track on the album. 1. Folder name: <path>/Madonna - Ray Of Light Tracks 1 -13 Title: (All 13 entered exactly as at: https://www.discogs.com/Madonna-Ray-Of-Light/release/1892372 Artist: Madonna (for all 13 tracks) Album Artist: Madonna (for all 13 tracks) Album: Ray of Light Genre: Pop Year: 1998 2. Folder name: <path>/Madonna - Confessions On A Dance Floor Tracks 1 -12 Title: (All 12 entered exactly as at https://www.discogs.com/Madonna-Confessions-On-A-Dance-Floor/release/690709 Artist: Madonna (for all 12 tracks) AlbumArtist: Madonna (for all 12 tracks) Album: Confessions On A Dance Floor Genre: Pop Year: 2005 Hope this helps. Many thanks. <edit> I should add tracks are named in following format: ArtistName - TrackTitle.ext e.g. Madonna - I Love New York.mp3 </edit> Edited July 26, 2019 by flakie
flakie 11 Posted July 26, 2019 Author Posted July 26, 2019 Checked modify date of all nfo files. The ones corresponding to the above 2 mentioned albums are the only ones updated today (for Madonna) so can be sure these were the 2 albums associated with the Lebanese Madonna.
ginjaninja 605 Posted July 26, 2019 Posted July 26, 2019 all the madonnas in these searches are the correct madonna. http://musicbrainz.org/ws/2/release/?query=release:ray%20of%20light%20AND%20artist:madonna http://musicbrainz.org/ws/2/release/?query=release:ray%20of%20light%20AND%20artist:madonna%20AND%20tracks:13 http://musicbrainz.org/ws/2/release/?query=release:ray%20of%20light%20AND%20artist:madonna%20AND%20tracks:13%20AND%20date:1998 search via release (Rather than release group) is good as this exposes other metadata to filter results and pick from results total tracks, year, are available Country of release and Medium of release, might be pretty to default to medium CD / local language when available and a match. http://musicbrainz.org/ws/2/recording/?query=artist:Madonna%20AND%20release:Ray%20of%20Light%20AND%20recording:Drowned%20World%20AND%20tnum:1%20AND%20tracks:13%20AND%20date:1998 searching by recording (trackname) and track number and release (album name) and tracks on release might be an additional cross check. its seems that a limited number (if necessary) of cross referenced lookups should achieve a single correct result. Even if 1 of track name+track number + total tracks artist album were missing or incorrect the musicbrainz api provides a way to achieve a high level of certainty of a match via some cross referenced lookups. if slowing down the matching due to multiple lookups ensures 100% accuracy then that sounds worth it.
flakie 11 Posted July 26, 2019 Author Posted July 26, 2019 (edited) OK thanks, so no explanation for why it did not work then. Maybe it was adding all 700+ albums at the same time? I found a quick way of fixing the duplicate problem so it is not too big a deal for me now. I had them all done in less than an hour. Only 30 or so duplicates from the 700+ albums (30+ of which are double albums). So a hit/accuracy rate of about 96%, which I guess is acceptable. Going to look at Picard to see what tags it adds and then try to script something to look at the nfo files and grab the MBID and add to mp3tag. Then add them manually myself for any new albums before adding to Emby. Edited July 26, 2019 by flakie
PenkethBoy 2068 Posted July 26, 2019 Posted July 26, 2019 Picard will add all the tags directly to the songs no need to script anything
ginjaninja 605 Posted July 26, 2019 Posted July 26, 2019 OK thanks, so no explanation for why it did not work then. Maybe it was adding all 700+ albums at the same time? I found a quick way of fixing the duplicate problem so it is not too big a deal for me now. I had them all done in less than an hour. Only 30 or so duplicates from the 700+ albums (30+ of which are double albums). So a hit/accuracy rate of about 96%, which I guess is acceptable. Going to look at Picard to see what tags it adds and then try to script something to look at the nfo files and grab the MBID and add to mp3tag. Then add them manually myself for any new albums before adding to Emby. it was less of an explanation of cause of your issue and more of a development suggestion to luke to try and make matching more robust. its seems a shame that artist matching is not as acurates as musicbrainz api searches suggest they could be: esp. when bios and images and other ids flow from that. didnt realise you had nfo files...dont know how they affect artist matching positively or negatively or even at all. i think you will find picard can detect and write out the tags to 700 albums pretty quickly...once you get used to remediating matches. i do all artists items at a time. a couple of mins per artist usually. surpised to hear you have artist and album artists music brainz ids in your nfos to import in any case. Sent from my SM-G955F using Tapatalk
flakie 11 Posted July 26, 2019 Author Posted July 26, 2019 Thanks I have installed Picard and am going to have a look soon. Sounds promising. I have Emby setup to create the nfo during the scan. They were not present until i added the music library to emby yesterday and did the first scan. I just did a random check of 30 or so nfo files and I would estimate that 70% have the MBIDs in them. Not sure why some do and others dont. It does not seem to have any bearing on which ones were wrongly detected artists or duplicates.
flakie 11 Posted July 26, 2019 Author Posted July 26, 2019 Picard looks a bit confusing a first glance. Will take more of a look when I have time to read up on it. just found another bad artist. Not a duplicate but incorrectly added with wrong MBID.It has picked up no artwork.3 associated albums. All are double albums.The tags are perfect for all 3 albums (no typos etc).The Rolling Stones.MusicBrainz Id:: a52940ca-3ec9-4a2b-89e4-ad358f64c8f7MusicBrainz URL: https://musicbrainz.org/artist/a52940ca-3ec9-4a2b-89e4-ad358f64c8f7Group: Andy Anderson & The Rolling StonesAnnotation:1950s US rockabilly band. (Nothing to do with the famous UK rock band.)Originally known simply as The Rolling Stones, see the explanation here.I added the correct MBID: b071f9fa-14b0-4217-8e97-eb41da73f598And all is good now. Artwork is added.
ginjaninja 605 Posted July 27, 2019 Posted July 27, 2019 (edited) Thanks I have installed Picard and am going to have a look soon. Sounds promising. I have Emby setup to create the nfo during the scan. They were not present until i added the music library to emby yesterday and did the first scan. I just did a random check of 30 or so nfo files and I would estimate that 70% have the MBIDs in them. Not sure why some do and others dont. It does not seem to have any bearing on which ones were wrongly detected artists or duplicates. if you are not invested in the nfos i would remove them and turn them off..at worst they just get in the way. i suspect they are more about exchanging metadata with 3rd party systems, in the past they have made library rebuilds less reliable, doesnt sound like u need them. with mo nfos you know its just about id3tags vs emby lookup logic. picard does have a steep learning curve. this may help add content to lhs cluster to put content into albums lookup on 1 or more clusters to search musicbrainz for matches drag track mismatches around and whole album mismatches back to lhs. right click cluster to manual search on mbz right click on matches to return alternative releases. once you have matching sussed, optionally investigate options re tagging, scripting, album art and moving content into a structure. i use a script to write out multiple album artists which i dont think Picard does by default. if you run and match ALL content via picard , i think you will find emby will be 100%. including your rolling stones example. if you run some content via picard your mileage may vary. Sent from my SM-G955F using Tapatalk Edited July 27, 2019 by ginjaninja
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now