SongKong Jaikoz

SongKong and Jaikoz Music Tagger Community Forum

Interactive tag from MusicBrainz:Open Discussion

I am just considering the best way to add an interactive tag from musicbrainz
in addition to the fully automated tag currently available, a number of people have requested such an option.

I would be interested in your ideas on how you would expect this to work, and what additional options you would like. Bear in mind that it has to work in the same way whether you are just tagging 10 tracks from the same album or 1000 tracks from different albums.

This is how I see it working:

The user selects ‘Manual Correct Tags from MusicBrainz’.
For each (selected)record
If record has puid we retrieve all tracks for that puid
If record has a title we do a metamatch and retrieve upto ten matches.
If we have processed ten records
Show Dialog for each record we list a table underneath showing
the artist/album/title/track no/score/puid for each potential match and
a button to allow this record to be used.

       The user reviews the possible matches, selects matches as 
       required  then dismisses the dialog.

       These records are updated.

       reset count.

Im not sure about how many records should be batched up, Ive chosen ten but I will make it user selectable. I could wait until the matching has completed but this would mean retaining all possible matches in memory which could cause resource problems, and I think it would make things difficult to manage.

This is a completely manual process in that no records will be updated unless the user selects them whereas the existing method is completely automated in that it will either match/not match based on the MusicBrainz Settings. Im not sure if users also want a semi-automated method that automatically matches records if they meet the score threshold but otherwise shows dialogs with possible options, in which case would you expect the dialog come up every so often as matching progresses or just once at the end.

and here a mockup of the dialog as I see it

Ive also taken Pitmans suggestion to put non -matching records aside, and
am going to implement as a number of filters similar to the duplicates filters, the filter will be:
Show records without Puid
Show records with Puid
Show records without MusicBrainzId
Show records with MusicBrainzId

Released today, as described above

Paul,

I like the possibility to manually update tags from Musicbrainz - great feature.

However, I personally find that the existing lookup design with using the song title as the only mbrainz search key does not quite meet my Jaikoz needs. :slight_smile:

I find that I often have a rather good idea of the name of the artist (give or take a few spelling errors), the album name (again, give or take a few minor string errors) and the track number (correct or missing). The song titles are sometimes correct, sometimes missing, and I do not trust them as an authority on identifying a song.

I would like to be able to query musicbrainz using a combination of these 3 fields (album, artist, track no) instead of just the title field as search key. I would also love for Jaikoz to, after sorting the spread sheet after sub folder, look vertically across neighbouring songs and assume that all the songs in the same folder belong to the same album by the same album artist and consider this when doing the musicbrainz lookup (and merge the lookup results accordingly).

If Jaikoz looked vertically within each subfolder, it could then possibly do away with those annoying musicbrain searches where 8 out of 10 songs in an album are matched and you have to manually edit the last 2, or where all songs are matched but to slightly different albums by the same artist.

I may be off on a tangent here and requesting something that is really hard to build, but think that it would be quite cool if it was possible.

Cheers from Sweden,
-A.

[quote=AndersL]I personally find that the existing lookup design with using the song title as the only mbrainz search key does not quite meet my Jaikoz needs. :slight_smile:
[/quote]
Not sure if I understand you correctly, are you saying you wish Jaikoz would use more than the title when trying to match tracks or something else? Because it already does, it uses the album,artist,track number and audio length where availble, but it is weighted towards the title as this is the item most likely to uniquely identify the track.

[quote=AndersL]
I would also love for Jaikoz to, after sorting the spread sheet after sub folder, look vertically across neighbouring songs and assume that all the songs in the same folder belong to the same album by the same album artist and consider this when doing the musicbrainz lookup (and merge the lookup results accordingly).

If Jaikoz looked vertically within each subfolder, it could then possibly do away with those annoying musicbrain searches where 8 out of 10 songs in an album are matched and you have to manually edit the last 2, or where all songs are matched but to slightly different albums by the same artist.

I may be off on a tangent here and requesting something that is really hard to build, but think that it would be quite cool if it was possible.

Cheers from Sweden,
-A.[/quote]
Once I have implemented support for MP4 I am going to add some features that try and minimize the number of albums songs are spread over. I may use some of your ideas above, but havent yet decided exactly what I am going to do.

Doh,

I stand corrected. I misunderstood your description of the lookup and assumed that it only checked using the track title. My bad.

Personally I would prefer the “one file at a time” approach I think Focher suggested in another thread I can’t find at the moment.
But if you stick to the current approach I would like to suggest the addition of a couple of options:

  • Make it possible to undo a selection. If you make an error, now the only way to correct it is to cancel the lookup and redo it. In other words click a line once selects it, a second click deselects it
  • Color coding differences so you can for example detect spelling errors even without selecting a suggested match. Which would be really great in combination with:
  • Placing a row of checkboxes beneath every column of every block of suggestions. The default settings for the checkboxes would be the selection made in the musicbrainz preferences. I hope you get what I mean. This would make it possible to customize the fields taken over from musicbrainz per song looked up.
    This can then be used for correcting spelling errors or for selecting just the fields you want to correct. I for example have a lot of tracks without an album. Now I can either do a lookup and select the most matching item which causes me to accept most likely wrong album names or I can first change the preferences select the tracks without albums and do the lookup. But would it not be nice to be able to do this on the fly?

On a sidenote, the automatic sizing of the window is not always right currently. If you lookup just 2 items sometimes the first is shown completely, the second only partly while there is ample room on the screen.

I think I covered my view on the Manual lookup option in the other thread along with a gui mockup so won’t rehash any of that other than to say that the main shortcoming I have found in the Manual Match process is that I often end up matching PUIDs from different release entries in the MB database.

Granted, a lot of times this is because there are duplicates in the MB database, but at least showing the MB Release Id would allow the user to easily match up to the same release entries.

The other thread, by the way, is at http://www.jthink.net/jaikozforum/posts/list/174.page.

[quote=ErwinB]Personally I would prefer the “one file at a time” approach I think Focher suggested in another thread I can’t find at the moment.
[/quote]
Maybe I need two Manual Corrects, a simple view and an advanced view with the advanced view only showing one file at a time because of the amount of space required to fit everything in. Im hesitant to do this by default as I think it will be too complicated for some people to understand, and I think one file at a time is slower.

Yep, that is needed

Thats an interesting idea that could work

Ok, I understand I can see the advantage of that , though the screen is getting cluttered now.

[quote=ErwinB]
On a sidenote, the automatic sizing of the window is not always right currently. If you lookup just 2 items sometimes the first is shown completely, the second only partly while there is ample room on the screen. [/quote]
OK.

Undo Selection, Display of Release id and Correct Window Size all implemented in Jaikoz 1.12 released today.