gsoc idea and query

Ian Monroe ian.monroe at
Thu Feb 26 00:51:36 UTC 2009

On Wed, Feb 25, 2009 at 5:45 AM, amit sethi <amit.pureenergy at> wrote:
> Hi ,
> I have been working on the idea of an automatic playlist generator .
> Basically what it does is
> Get the metadata from the music file.Then using API it will get the
> list of similar songs , artist , genre .
> Now this set would be compared to other files in your
> (filesystem/folder/playlist) so
> for song A ,B:
> Artist_Similarity_score(A,B)=length of (set_Artist_similar(A) interjection
> set_Artist_similar(B)set_Artist_similar(A) union set_Artist_similar(B))
> These scores are clustered by eucledian distance.
> The problem with this is that it is relatively  slow but i was thinking if
> it can be in some manner tweaked and used as scoring system for Amarok . I
> would like to do that as a gsoc project.
> Also I have seen a few ideas at
> and i
> would like to query about them :
> 1.Project: Analyzer Support mentor Ian Monroe
> Although I do not have exactly the required experience i have worked  with
> pymedia and Snack toolkit.
> and also used the FFT library in python .I have no experience working in C++
> but i use python in object oriented way and have learned C++ in school. I
> would like to know the basic idea of how the project needs to be implemented
> and the major tools that one should be acquainted with to see if i can
> implement it.

Well like the project idea says, you add the API to Phonon (this kind
of already exists in an experimental namespace I think, but probably
needs to be cleaned up and such), and then with the way Phonon works
you actually implement the API in the backends.

And you add analyzer support in an application like Dragon Player, JuK
or Amarok so that you can test the API and the backend.

See the response I gave Ricard for more info. There's several
different ways you could go with this idea I think.


More information about the Amarok mailing list