Re: [Rhythmbox-devel] In the Mood plugin thoughts

[Neil Loknath <neil loknath gmail com>]

Good to see that there is some interest in this type of feature!

I'm going to take a look at autoqueue and I'm in the process of
compiling Banshee and the Mirage plugin to test that out.

I'm not too keen on using last.fm as a source for retrieving music
similarity data, as a lot of the music it would recommend would probably
not even exist in my library.  Nevertheless, I won't totally dismiss it
until I try it.  I'm looking forward to testing it out.

I did some snooping around, and it seems that there hasn't been much
development around playlist generation based on tonality.  Most of the
work is based on timbre.  If anyone could shed some light on the reasons
for that, I'm all ears.  Anyway, I think that a playlist generator based
on tonality and tempo would be really cool.  I will have to continue
researching to determine the feasibility, however.  I know the Serato
Scratch DJ software implements some reliable BPM and key analysis
features, so it must be possible.

Neil

On Wed, 2009-02-04 at 14:05 +0100, eric casteleijn wrote:
> Alex Bennee wrote:
> > On Wed, Feb 4, 2009 at 6:20 AM, Neil Loknath <neil loknath gmail com> wrote:
> >> I recently compiled and installed Charlotte Curtis' GSOC project form
> >> last year: the In the Mood plugin, which can be found at
> >> http://code.google.com/p/rhythmbox-predictive-playback/downloads/list
> >>
> >> Congrats on a very cool project, Charlotte!  I have been using it
> >> intensely since installing it, and I quite like the concept of it.  It
> >> seems that intelligent playlists are quite popular right now (ie. iTunes
> >> Genius feature).  And, I think the trend will continue.  With that being
> >> said, I have some thoughts on expanding on Charlotte's work.  In fact, I
> >> am hoping that Charlotte will chime in to this and join the discussion.
> > 
> > I'm currently looking at the possibility of doing something similar
> > but using the last.fm web API to generate potential matches (which is
> > how I think Genius works on iTunes). It will be interesting to see how
> > the two compare in performance.
> > 
> > I have no code to at the moment as I'm just familiarising myself with
> > writing Python plugins for Rhythmbox. However I shall post as soon as
> > I have something worth sharing.
> 
> Please have a look at autoqueue[1] in that case: it's a plugin that
> already does this, for a number of players, among which Rhythmbox. I've
> also ported the mirage acoustic similarity plugin that Banshee has from
> C# to python, so in fact, it can do both the last.fm lookup *and*
> something very similar to Charlotte Curtis' plugin. (The different
> similarities can be used together or separate, and weighted to taste.)
> 
> The rhythmbox plugin may not be 100% up to date with the latest changes,
> I will check if I can bring it up to date after work tonight.
> 
> Of course developing your own can be a great experience too, so I'm not
> saying you shouldn't, but I like collaborating where feasible. ;)
> 
> Re: Genius in iTunes: People I know who use iTunes seem to agree the
> recommendations there are pretty shallow and silly sometimes. I think
> the last.fm ones are better, and the acoustic similarity rocks even
> more, since it gives equal time to completely unknown artists, and
> completely disregards genres and other artifacts. (In last.fm a lot of
> artists/tracks are similar because they are from the same country, but
> share nothing else in common. Understandable because of the way they
> gather their data, but acoustic analysis completely fixes this effect.)
> 
> [1] http://code.google.com/p/autoqueue/
>