Luminal.org - Next Generation

IMMS The Next Generation

     or Where do we go from here?

WARNING: very out of date

Here are some long term ideas for the IMMS project. I don't know how many of these I will actually get around to working on, and certanly not when, but if you think that something on this list is a good (or bad!) idea, say so! I will certainly take your feedback into consideration.

Better song identification/centralized metadata repository

IMMS already has a pretty good system for guessing the artist / title of a song, I could use that. But what I would really like to do is to create a free (and better!) alternative to MusicBrainz. I am not exactly sure how I want to do this (yet). If you have ideas on implementing signal identification/ acoustic signature algorithms, be sure to let me know :)

Actually, here's my secret plan:

Make IMMS popular
Design an algorithm for signature calculation
Allow IMMS users to submit statistics for their music collection
Use the collected data to refine the algorithm
???
Profit!!

In other news.... Is MusicBrainz actually as Evil as I think? Who thinks we should not waste time on this and just use it as is? And anyway, I am really reluctunt to rely on any system like that too heavily, because it ties IMMS into depending on a network connection.

Actually, musicbrainz appears free now.
- Free as in beer, but not free as in speech. --mag
  - I just checked today (2006-01-31) and their page says the SDK is licensed under the LGPL.
    - Client libs are LGPL, core of the system - the fingerprint server - is closed source proprietary --mag

Something cool and distributed, a la gnod, etc

Well there're all sorts of things that can be done here, but they all sort of depend on having Better song identification

Take over the world. Or at least XMMS's playlist.

XMMS kinda sucks. Maybe I should write a better player all together. Well maybe not all at once. For example start by writing a playlist editor replacement (basically a "proxy" player that uses XMMS to do the actual playing). This would allow us to work around most of the XMMS plugin interface restrictions and let IMMS work to its fullest potential.

Please don't try to establish yet-another-player. Think of all the infrastructure of input, output and other plugins that XMMS has already. I think a playlist replacement would be the best. Does this still bind you to the plugin interface? How did qbble solve that?

TODO: finish writing this