Just needed to copy my Rhythmbox ratings from one machine to another. So, a quick script to do it: rhythmbox-ratings-export.py. It assumes that all your mp3s are at the same path on both machines, which is fine for me because they are (my music folder is in Ubuntu One), and it’s basically very stupid and venial and you shouldn’t do it this way. But it worked when I needed to quickly transfer a bunch of ratings and was prepared to ponce about on the command line, so someone else might find it useful.
Memo to self: write plugin to sync this stuff properly.
Yes! Please write a plugin to sync this properly.
It could even sync to Ubuntu One, then it should probably be on a program agnostic format so that Banshee also can read/write the ratings. (Number of times played as well).
Dude yes, we need this badly. I’ve been wishing for an app that can magically sync and merge my rhythmbox song ratings between my two machines for ages. Manually keeping two collections of different sizes (one is a subset of the other) in sync while one of them has 13 000 songs is not exactly trivial to do by hand.
I made you a tiny little pony: http://git.gnome.org/browse/rhythmbox/commit/?id=2e109b246ec694774ec38ff0e013823690d12264
Ha! Nice one Jonathan! thanks
I have every intention of writing a proper plugin that syncs to Ubuntu One in a program agnostic format
That’s not a quick hack, though. The biggest problem is having a unique identifier for a song…hopefully musicbrainz IDs will be it…
Yeah, I’m tired of having parts of my music spread across my home desktop, work desktop, laptop, and N900. I’ve been experimenting with migrating my old DAD project to couchdb, so that it will handle both the audiofiles themselves being duplicated across machines, as well as things like ratings.
musicbrainz ID’s is one way to go, but what about tracks that aren’t (and shouldn’t be) in musicbrainz, like live bootlegs for example ? An acoustic fingerprint will be better…
My solution probably won’t be directly applicable, because instead of just identifying audiofiles as the basic concept, I work from something called Track which is represented by a Slice of an AudioFile. For example, ‘All apologies’ from Nirvana being a Track, represented by the first slice of 4 minutes of sound on the In Utero album’s corresponding track number, without the twenty minutes of silence and bonus track after it.
A track can have multiple slices into multiple audiofiles, and my player would select the best quality for the current device.
Anyways, pie-in-the-sky stuff
Yeah, I’m tired of having parts of my music spread across my home desktop, work desktop, laptop, and N900. I’ve been experimenting with migrating my old DAD project to couchdb, so that it will handle both the audiofiles themselves being duplicated across machines, as well as things like ratings.
musicbrainz ID’s is one way to go, but what about tracks that aren’t (and shouldn’t be) in musicbrainz, like live bootlegs for example ? An acoustic fingerprint will be better…
My solution probably won’t be directly applicable, because instead of just identifying audiofiles as the basic concept, I work from something called Track which is represented by a Slice of an AudioFile. For example, ‘All apologies’ from Nirvana being a Track, represented by the first slice of 4 minutes of sound on the In Utero album’s corresponding track number, without the twenty minutes of silence and bonus track after it.
A track can have multiple slices into multiple audiofiles, and my player would select the best quality for the current device.
Anyways, pie-in-the-sky stuff
Sorry for the double post – clicking post comment gives no visual feedback that it works. Also for some reason the Name/website/comment fields look huge with huge fonts for me… Using Firefox here.
Yesterday I wrote a rhythmbox plugin that uploads the ratings for tracks with musicbrainz-UUID to the website: https://github.com/funkycoder/mbratings (halfway through I thought to check the API and it doesn’t actually allow bulk downloading the user’s ratings to import on another computer, and checking every single track in the library won’t work because it takes 1sec each, but I opened a bug with them, maybe there will be a way. Or HTML scraping)