Post 2.0.0 release plan discussion

Daniel Jones danielcjones at gmail.com
Mon Nov 24 23:27:06 CET 2008


I'm not sure if next quarter will be as busy as this one, but if/when
I have time, I'd be very interested to build on the basis I laid for
dynamic playlists.

Soren's aucoustic fingerprinting is on very exciting idea that will
integrate well.

Others biases that should be written are

Last.fm biases:
Similar artists is the obvious one. This would allow you to seed the
playlist with a track you like, then each track that gets added is
similar to the last. Kind of like last.fm radio streams, but just with
your local collection.

There probably other good uses for last.fm statistics that I haven't thought of.


Ordering biases:
A bias that will try to produce tracks that, for example have a
greater length than the last. This probably has pretty limited use,
but someone might enjoy it.
I started writing this a while back, but never found time to finished it.


Also, I want to continue to experiment with improving the algorithm,
since this is really interesting problem to apply the latest and
greatest optimization and search techniques to.

There are likely some minor improvements that could be made to the ui
to make it more useable. For example, it is not completely obvious
when the dynamic playlist is turned on (you have to have that panel
open and see the button depressed). It would also be good to see some
kind of message in the bias list explaining what biases are and why
you want to add one.

-Daniel


More information about the Amarok-devel mailing list