[mrob]

Okay, I PM'd you a link to the file. Meanwhile, I tried retagging the files with Picard set to ID3 v2.4 format and that seems to work.

I can't really see the difference in Picard's display of the tags, but there must be something different happening in there.

Thanks again for all your help with this. If you need any more info about the crash, just let me know.

[DaveBlake]

Yes "something different happening in there", got your files and reproduced cash during scan. Now to investigate

First thing I notice is that the files also have ID3 v1 format tags (which my test files didn't), that makes a good place to start looking. I'll report what I find, and hopefully will be able to make the scanning process more robust.

[mrob]

Yes, it seems like Picard sets up Tag 1 with ID3 v1.1. Since it duplicates Tag 2 and just takes up a little more space, I would prefer no Tag 1.

Maybe there's a setting in Picard to turn this off, but I haven't noticed it.

EDIT: I looked at Picard again and there's an obviously-named setting Options > Tags > Also include ID3v1 tags in the files. I guess it was enabled by default.

[DaveBlake]

My guess was wrong, the issue is not the presence of ID3 v1 tags, it is that the ALBUMARTISTS tag is missing and the artists and artist mbids values ("Simonetti", "Morante", "Pignatelli") are in a different order to the album artists and album artist mbids ("Simonetti", "Pignatelli",  "Morante").

It needs a settings chnage in Picard to get the ALBUMARTISTS tag needed to support albums with multiple album artist, sadly it does not do that by default. To achieve that modiy the script settings (in Picard), found under Options>Options>Advanced>Scripting, add
$setmulti(albumartists,%_albumartists%)

Of course I also have a Kodi fix to do prevent the crash, as this example has usefully revealed a bug in the extra processing it does when there is a mis-match in number of artist names to mbid tags. With the separate frames in v2.4 format tags it does not have such a mis-match.

EDIT: PR raised https://github.com/xbmc/xbmc/pull/15635

[mrob]

Thanks much!!