-
Notifications
You must be signed in to change notification settings - Fork 1.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
chroma: Bias toward looking up more relevant releases using date/country #3017
Comments
Hi! This sounds like an interesting proposal. I think I might be misunderstanding one thing, though: in the importer interface, we typically sort releases by their similarity (i.e., the most "likely" match goes first). In what context are you talking about sorting stuff? |
The Acoustid lookup returns a list of recordings with sublists of releases which contain the MusicBrainz ID and a few other details including the date and country Acoustid API (The request the plugin uses is recordings+releases). The current behaviour (on line 190) uses the prefix function to just take the first five releases in the list from the Acoustid lookup then look them up in MusicBrainz and return those options, but since the order of the list seems to be semi random (possibly based on date submitted or something) it basically returns random albums that contain the song/s, which a lot of the time are best of or various artist albums. This would sort the full list from Acoustid first (before the prefix and MusicBrainz lookups occur) to hopefully provide more useful options. And since the whole point of this plugin is to get the info from the fingerprint rather than the metadata in the file, I don't think there are many other ways of selecting relevant results. And beets still gives a few options based on any tags the file may have. |
Got it; thanks! That helps a lot. So the goal is to make some guesses about which releases are likely to be the most "productive" ones to look up. That sounds great! If we do implement this, it would be nice to have some side-by-side comparisons where matching failed (or went poorly) before the change but went well after the change. |
Same settings used for all:
I removed all the tags from the files to show that the results are just from the fingerprinting Tom's Diner by Suzanne Vega:Old:
New:
Mr Blue Sky by Electric Light Orchestra:Old:
New:
|
Cool! Would you mind opening a pull request with those changes so we can discuss them? |
By default, the releases provided by the
chroma
plugin are effectively random, because it just presents the first five that are in the list supplied by Acoustid.I would like to suggest a change that first sorts them according to the date of release and then their country depending on the
match
configuration options. The default config will still result in the same behaviour as before.Archer4499/beets@7a8f980
For example the following configuration:
will sort by date (oldest first) then by the preferred countries (US then GB/UK then others).
I find this useful because when I have just one or two songs from an artist I like tagging them with the original album that they were released in.
The main point of this being posted as an issue was to check if I should just make it sort by date by default (and still doing countries as an option), because I think it picking random releases isn't actually useful.
There is also possibly the separate issue of the mb.py (Line 119) file not using regex entries in the
countries
entry.The text was updated successfully, but these errors were encountered: