Incorrect song matching

[mgray88]

Problem

Running beet import it wasn't able to identify the album name from the folder name for search. So I entered the MusicBrainz id that I wanted to match.

Led to this problem:

That worked, and matched correctly with the first two songs as can be seen; however, it incorrectly tried to match the third song with a different song

Setup

• OS: ubuntu 17.04
• Python version: 2.7.13
• beets version: 1.3.18
• Turning off plugins made problem go away (yes/no): didn't try, only plugin is discogs, which didn't seem to cause the problem

My configuration (output of beet config) is:

directory: /mnt/Media/Music
import:
    move: yes
plugins: discogs
discogs:
    user_token: ***
paths:
    singleton: $albumartist/$album%aunique{}/$track $title
match:
    preferred:
        countries: ['US', 'GB|UK']
        media: ['CD', 'Digital Media|File']

[sampsyo]

Interesting! Our matching heuristic isn't perfect, of course. If you're interested in digging into exactly why this didn't match the way you expected, please provide a test case we can reproduce.

[jackwilsdon]

This sounds like we should maybe consider non-alphanumeric changes as being "closer" than changing characters, maybe weighting them differently? (I'm not quite sure how our distance calculations work now though).

If anyone is interested in taking a look, the area of interest seems to be the string_dist function (although correct me if I'm wrong @sampsyo).

[sampsyo]

Yes, that’s the right function to look at! I don’t quite understand the hypothesis, though, @jackwilsdon—which non-alphanumeric characters are you referring to?

[jackwilsdon]

It seems to have ranked "Ride the River" higher than "My Father's Eyes" when searching for "My Fathers Eyes" :confused:

[sampsyo]

Got it! So the hypothesis is that the apostrophe is counting for more than the letters. I can’t think of anything specific that would cause this, but it’s worth investigating.

(Sorry for the close/reopen noise; I hit the wrong button. 🙄)

[jackwilsdon]

Just did a little bit of testing and got these results;

str1	str2	string_dist
My Fathers Eyes	Ride the River	0.6153846153846154
My Fathers Eyes	My Father's Eyes	0

So I guess that proves my hypothesis wrong!

[sampsyo]

Hmm... thank you for investigating! This is certainly mysterious.

[nbartnik]

Hi,

This issue occurs when the following conditions are met: 1) Track title alpha-numeric sort order differs from the album track list order.
2) Track number is missing from the tags

e.g

Poor matching with only track title, album title and artist:

Tagging:
    Jonathan Coulton - Artificial Heart
URL:
    https://musicbrainz.org/release/6f238401-4ba3-4691-8c62-02242627a28c
(Similarity: 48.6%) (tracks) (CD, 2011, US, Jocoserious Records)
 * Down Today.mp3 (#0)               -> Sticking It to Myself (#1) (title)
 * Artificial Heart.mp3 (#0)         -> Artificial Heart (#2) (title)
 * Dissolve.mp3 (#0)                 -> Nemeses (#3) (title)
 * Alone at Home.mp3 (#0)            -> The World Belongs to You (#4) (title)
 * Fraud.mp3 (#0)                    -> Today With Your Wife (#5) (title)
 * Sucker Punch.mp3 (#0)             -> Sucker Punch (#6) (title)
 * Glasses.mp3 (#0)                  -> Glasses (#7) (title)
 * Good Morning Tucson.mp3 (#0)      -> Je suis Rick Springfield (#8) (title)
 * The World Belongs to You.mp3 (#0) -> Alone at Home (#9) (title)
 * Nemeses.mp3 (#0)                  -> Fraud (#10) (title)
 * Je suis Rick Springfield.mp3 (#0) -> Good Morning Tucson (#11) (title)
 * Now I Am an Arsonist.mp3 (#0)     -> Now I Am an Arsonist (#12) (title)
 * Nobody Loves You Like Me.mp3 (#0) -> Down Today (#13) (title)
 * The Stache.mp3 (#0)               -> Dissolve (#14) (title)
 * Sticking It to Myself.mp3 (#0)    -> Nobody Loves You Like Me (#15) (title)
 * Still Alive.mp3 (#0)              -> Still Alive (#16) (title)
 * Want You Gone.mp3 (#0)            -> Want You Gone (#17) (title)
 * Today With Your Wife.mp3 (#0)     -> The Stache (#18) (title)

Same album, but with track name sort order matching the album tracklist order:

Tagging:
    Jonathan Coulton - Artificial Heart
URL:
    https://musicbrainz.org/release/6f238401-4ba3-4691-8c62-02242627a28c
(Similarity: 92.2%) (tracks) (CD, 2011, US, Jocoserious Records)
 * aa Sticking It to Myself (#0)    -> Sticking It to Myself (#1) (title)
 * ab Artificial Heart (#0)         -> Artificial Heart (#2) (title)
 * ac Nemeses (#0)                  -> Nemeses (#3) (title)
 * ad The World Belongs to You (#0) -> The World Belongs to You (#4) (title)
 * ae Today With Your Wife (#0)     -> Today With Your Wife (#5) (title)
 * af Sucker Punch (#0)             -> Sucker Punch (#6) (title)
 * ag Glasses (#0)                  -> Glasses (#7) (title)
 * ah Je suis Rick Springfield (#0) -> Je suis Rick Springfield (#8) (title)
 * ai Alone at Home (#0)            -> Alone at Home (#9) (title)
 * aj Fraud (#0)                    -> Fraud (#10) (title)
 * ak Good Morning Tucson (#0)      -> Good Morning Tucson (#11) (title)
 * al Now I Am an Arsonist (#0)     -> Now I Am an Arsonist (#12) (title)
 * am Down Today (#0)               -> Down Today (#13) (title)
 * an Dissolve (#0)                 -> Dissolve (#14) (title)
 * ao Nobody Loves You Like Me (#0) -> Nobody Loves You Like Me (#15) (title)
 * ap Still Alive (#0)              -> Still Alive (#16) (title)
 * aq Want You Gone (#0)            -> Want You Gone (#17) (title)
 * ar The Stache (#0)               -> The Stache (#18) (title)

[sampsyo]

Hi! I think the problem you're seeing here, @nbartnik, is that the first version has no title tags. You're seeing the filenames as a fallback. Please consider enabling the fromfilename plugin.

[stale[bot]]

Is this still relevant? If so, what is blocking it? Is there anything you can do to help move it forward?

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[MCMXD]

Are there any add ons to correct this sort of error via CLI during import?