Core ideas behind this project:

* Your filesystem should remain tidy and clean – let’s pack tracks into uncompressed archives a.k.a. “libraries” (this was inspired by a common practice of video game developers to store all game resources in one or several packages). Less files – less clutter.

* Tags suck (especially ID3) – let’s utilize XML/YAML/JSON to describe entire albums.

* Storing cover image in every track is a waste of space – let’s have one TIFF/PNG/FLIF file per library.

* Being able to verify integrity of files in your music collection is a good feature, so let’s use hash lists (MD5, SHA1 etc.) for libraries.

* Rather than having to dig through lots of directories in order to select what you want to listen at the very particular moment it is much more convenient to have a centralized per-user database with a quick access to any track/album in your precious collection.

sourceforge.net/projects/arita/